Lionnie, Regina (2017) ANDROGENIC HAIR PATTERN RECOGNITION FOR BIOMETRIC IDENTIFICATION. S2 thesis, Universitas Mercu Buana Jakarta-Menteng.
|
Text (Cover)
hal cover.pdf Download (1MB) | Preview |
|
|
Text (Abstrak)
abstrak.pdf Download (477kB) | Preview |
|
![]() |
Text (Bab 1)
BAB 1.pdf Restricted to Registered users only Download (360kB) |
|
![]() |
Text (Bab 2)
BAB 2.pdf Restricted to Registered users only Download (1MB) |
|
![]() |
Text (Bab 3)
BAB 3.pdf Restricted to Registered users only Download (820kB) |
|
![]() |
Text (Bab 4)
BAB 4.pdf Restricted to Registered users only Download (1MB) |
|
![]() |
Text (Bab 5)
BAB 5.pdf Restricted to Registered users only Download (274kB) |
|
![]() |
Text (Daftar Pustaka)
Hal daftar pustaka dan lampiran.pdf Restricted to Registered users only Download (1MB) |
Abstract
To identify criminal or pedophile in online child pornography images and video is a challenging task when the faces and other distinguish features are not shown. To address these kind of problems, the system of recognition using androgenic hair pattern is being developed. Androgenic hair pattern stated to be the newest biometric trait since 2014. There were 4 different methods that were compared in this thesis, Haar wavelet transform, principal component analysis, hierarchical Gaussian scale space and scale invariant feature transform (SIFT). For the first three methods, 400 images were tested for the recognition system. Hierarchical Gaussian scale space produced 94,23 % precision of recognition using 10-fold cross validation with histogram equalization. The method defeated other two methods, Haar wavelet transform with 83,48 % precision of recognition and principal component analysis with 75.19 % precision of recognition. The next design used smaller version of androgenic hair database, 50 images with extreme condition of only 2 images variation for one respondent. For this latest system design, the SIFT algorithm gave the best performance of 38% precision of recognition and was superior to other previous three methods, Haar wavelet transform, principal component analysis and hierarchical Gaussian scale space that only produced around 30-32 % precision of recognition. Keywords: androgenic hair pattern; biometric identification; Haar wavelet transform; principal component analysis; scale-invariant feature transform. Untuk dapat melakukan identifikasi pelaku kriminal atau pedofil pada kasus kejahatan seksual anak-anak yang tertangkap kamera atau video digital menjadi tantangan yang sulit ketika wajah atau fitur unik dari kriminal tersebut tidak terlihat. Untuk dapat mengatasi kesulitan ini, telah dibangun sistem pengenalan berbasis pola rambut androgenik. Pola rambut androgenik menjadi ciri biometrik baru sejak tahun 2014. Terdapat 4 metode yang dibandingkan di dalam tesis ini, transformasi wavelet Haar, analisis komponen utama, skala ruang hierarki Gauss dan scale- invariant feature transform (SIFT). Untuk ketiga metode pertama, 400 gambar digital diuji untuk sistem pengenalan. Metode skala ruang Hierarki Gauss menghasilkan keakuratan presisi sistem terbaik yaitu 94,23 % menggunakan validasi silang 10-lipat dengan ekualisasi histogram. Metode ini mengalahkan transformasi wavelet Haar yang memberikan presisi 83,48 % dan analisis komponen utama yang menghasilkan 75,19 % presisi keakuratan. Desain sistem selanjutnya menggunakan versi basis data yang lebih kecil, 50 gambar digital dengan kondisi ekstrem hanya 2 variasi gambar yang diambil dari setiap responden. Untuk desain sistem terakhir ini, algoritma SIFT memberikan keakuratan presisi terbaik yaitu 38 % dan mengalahkan ketiga metode sebelumnya transformasi wavelet Haar, analisis komponen utama dan skala ruang Hierarki Gauss yang hanya menghasilkan sekitar 30-32% presisi keakuratan. Kata kunci: analisis komponen utama; identifikasi biometrik; pola rambut androgenik; scale-invariant feature transform; transformasi wavelet Haar.
Item Type: | Thesis (S2) |
---|---|
NIM/NIDN Creators: | 55415120017 |
Uncontrolled Keywords: | Keywords: androgenic hair pattern; biometric identification; Haar wavelet transform; principal component analysis; scale-invariant feature transform. Kata kunci: analisis komponen utama; identifikasi biometrik; pola rambut androgenik; scale-invariant feature transform; transformasi wavelet Haar. |
Subjects: | 600 Technology/Teknologi > 620 Engineering and Applied Operations/Ilmu Teknik dan operasi Terapan > 621 Applied Physics/Fisika terapan |
Divisions: | Pascasarjana > Magister Teknik Elektro |
Depositing User: | SITI NOVI NUR CAHYANI |
Date Deposited: | 28 Sep 2022 08:09 |
Last Modified: | 28 Sep 2022 08:09 |
URI: | http://repository.mercubuana.ac.id/id/eprint/69666 |
Actions (login required)
![]() |
View Item |