Khotimah, Mutiara Husnul (2026) OPTIMALISASI METODE ORIENTED FAST AND ROTATED BRIEF UNTUK PENGENALAN WAJAH MENGGUNAKAN RANDOM SAMPLE CONSENSUS DAN HAMMING DISTANCE. Skripsi (Bachelor) thesis, Universitas Muhammadiyah Bengkulu.
![[img]](http://repository.umb.ac.id/style/images/fileicons/text.png) |
Text (Fulltext)
MUTIARA HUSNUL KHOTIMAH_2255201092_fulltext.pdf - Submitted Version Restricted to Repository staff only Download (2MB) | Request a copy |
|
Text (Turnitin)
MUTIARA HUSNUL KHOTIMAH_2255201092_Turnitin.pdf - Submitted Version Restricted to Repository staff only Download (11MB) | Request a copy |
|
Text (ABSTRAK + BAB 1)
MUTIARA HUSNUL KHOTIMAH_2255201092_ABSTRAK + BAB 1.pdf - Submitted Version Download (629kB) |
|
Text (BAB 2)
MUTIARA HUSNUL KHOTIMAH_2255201092_BAB 2.pdf - Submitted Version Download (606kB) |
|
Text (BAB 3)
MUTIARA HUSNUL KHOTIMAH_2255201092_BAB 3.pdf - Submitted Version Restricted to Repository staff only Download (744kB) | Request a copy |
|
Text (BAB 4)
MUTIARA HUSNUL KHOTIMAH_2255201092_BAB 4.pdf - Submitted Version Restricted to Repository staff only Download (1MB) | Request a copy |
|
Text (BAB 5)
MUTIARA HUSNUL KHOTIMAH_2255201092_BAB 5.pdf - Submitted Version Download (231kB) |
|
Text (Ref + Lamp)
MUTIARA HUSNUL KHOTIMAH_2255201092_Ref + Lamp.pdf - Submitted Version Download (569kB) |
Abstract
Sistem pengenalan wajah berbasis ekstraksi fitur lokal masih menghadapi kendala berupa tingginya kecocokan fitur yang keliru (outlier) pada tahap pencocokan deskriptor, sehingga akurasi identifikasi tidak stabil pada variasi pencahayaan dan pose wajah. Penelitian ini bertujuan mengimplementasikan metode Oriented FAST and Rotated BRIEF yang dioptimalkan dengan menggabungkan Random Sample Consensus sebagai tahap verifikasi untuk menyaring pasangan fitur outlier, serta Hamming Distance sebagai metrik pengukuran kemiripan antar deskriptor biner. Penelitian menggunakan metode eksperimen kuantitatif: keypoint dan deskriptor diekstraksi dari citra wajah, dicocokkan berdasarkan Hamming Distance, kemudian disaring menggunakan RANSAC agar hanya pasangan fitur yang konsisten secara geometris yang digunakan. Sistem diimplementasikan menggunakan Python dan diuji melalui skenario identifikasi satu-ke-banyak terhadap 1000 data uji. Hasil pengujian menunjukkan accuracy 91,0%, precision 91,51%, recall 91,51%, dan F1-score 91,51% (TP=485, TN=425, FP=45, FN=45), menunjukkan RANSAC meningkatkan konsistensi hasil pencocokan Hamming Distance sehingga kesalahan identifikasi dapat ditekan. Dengan demikian, optimalisasi ORB melalui kombinasi RANSAC dan Hamming Distance menghasilkan sistem pengenalan wajah yang lebih akurat, efisien, dan stabil. Face recognition systems based on local feature extraction still face the challenge of a high rate of feature mismatches (outliers)during the descriptor matching stage, resulting in unstable identification accuracy under varying lighting conditions and facial poses. This study aims to implement an optimized version of the Oriented FAST and Rotated BRIEF method by combining random sample consensus as a verification stage to filter out outlier feature pairs, and hamming distance as a metric for measuring similarity between binary descriptors. This study employed a quantitative experimental method: keypoints and descriptors were extracted from facial images, matched based on hamming distance, and then filtered using RANSAC to ensure that only geometrically consistent feature pairs were utilized. The system was implemented using python and tested via a one-to-many identification scenario on 1.000 test data points. The test result showed an accuracy of 91.0%, precision of 91,51%, recall of 91,51%, and F1-score of 91,51% (TP=485, TN=425, FP=45, FN=45). The result indicate that RANSAC improves the consistency of hamming distance matching result, thereby reducing identification errors. Consequently, the optimisation of ORB through a combination of RANSAC and hamming distance yields a more accurate, efficient, and stable facial recognition system.
| Item Type: | Thesis (Skripsi (Bachelor)) |
|---|---|
| Additional Information: | Pembimbing : Dr. Marissa Utami, S.Kom., M.Kom |
| Uncontrolled Keywords: | Ekstraksi Fitur Lokal, Deskriptor Biner, Verifikasi Geometris, Outlier, Confusion Matrix |
| Subjects: | Universitas Muhammadiyah Bengkulu > 02-Fakultas Teknik > 55201-(S1) Teknik Informatika 02-Fakultas Teknik > 55201-(S1) Teknik Informatika |
| Divisions: | 02-Fakultas Teknik > 55201-(S1) Teknik Informatika Subjek Terkait > 02-Fakultas Teknik > 55201-(S1) Teknik Informatika |
| Depositing User: | Mrs Meri Susanti R |
| Date Deposited: | 14 Aug 2026 03:23 |
| Last Modified: | 14 Aug 2026 03:23 |
| URI: | http://repository.umb.ac.id/id/eprint/2971 |
Actions (login required)
 |
View Item |