Aini, Lusiana Nurul (2017) Studi Kinerja Metode Ekstraksi Fitur Local Line Binary Pattern dan Scale Infariant Feature Transform pada Aplikasi Palm dan Finger Vein Recognition. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
5113100147-Undergraduate-Theses.pdf - Published Version Download (5MB) | Preview |
Abstract
Dalam perkembangan teknologi, sosial, maupun ekonomi saat ini validasi untuk keamanan informasi telah meningkat. Metode otentikasi telah berkembang, seperti berbasis karakteristik biometrik yang tidak mungkin dilupakan maupun dipalsu yaitu metode otentikasi biometrik menggunakan teknik pengenalan pola pembuluh darah. Untuk melakukan pengenalan pembuluh darah, akan dilakukan proses ekstraksi fitur. Proses ekstraksi fitur ini sangat berpengaruh dalam tahap pengenalan pola setiap individu.
Tugas akhir ini mengusulkan perbandingan metode ekstraksi fitur menggunakan metode Local Line Binary Pattern (LLBP) dan Scale Invariant Feature Transform (SIFT). Tujuan dari tugas akhir ini yaitu untuk mengetahui performa antara dua metode tersebut dalam melakukan pengenalan pola pembuluh darah pada palm maupun finger vein. Data masukan berupa data testing dan data training palm dan finger vein. Proses pengenalan pola pembuluh darah terdapat 3 proses utama yaitu preprocessing yang bertujuan untuk memperbaiki citra agar terlihat pola pembuluh darah, kemudian ekstraksi fitur untuk mendapatkan fitur dari pola pembuluh darah, dan tahap terakhir yaitu melakukan similarity. Sehingga akan mendapatkan hasil akurasi setiap metode yang diimplementasikan pada setiap dataset.
Data yang digunakan untuk uji coba yaitu 35 individu data hasil scan palm dan finger vein. Untuk rata-rata hasil akurasi uji coba pengenalan palm vein menggunakan metode ekstraksi fitur LLBP yaitu 85.72%. Sedangkan menggunakan metode SIFT adalah 58.93%. Untuk pengenalan finger vein menggunakan metode LLBP yaitu 56.57%. Sedangkan menggunakan metode SIFT mempunyai rata-rata akurasi sebesar 80.72%. Dari hasil uji coba dapat disimpulkan bahwa metode LLBP lebih cocok digunakan untuk pengenalan palm vein, sedangkan metode SIFT lebih cocok digunakan untuk pengenalan finger vein.
=====================================================================
In current technological, social, and economic developments the validation for
information security has
increased. However, traditional authentication methods such as
passwords and keys can not meet expectations because they are
vulnerable to being forgotten, lost or stolen. Given this, it will be
able to identify and authenticate b
ased on biometric
characteristics that can not be forgotten or falsified, namely
biometric authentication method using vein pattern recognition
technique on fingers (Finger Vein) and palms (Palm Vein). To
perform the introduction of individual blood vessel
s, a feature
extraction process or characteristic of the blood vessels will be
performed. This feature extraction process is very influential in the
pattern recognition stage of each individual.
In this final project proposes to compare feature extraction
method using Local Line Binary Pattern (LLBP) method and Scale
Invariant Feature Transform (SIFT). The purpose of this final task
is to determine the performance between the two methods in
performing pattern recognition of blood vessels in the palm and
fin
ger vein. The input data are data testing from palm and finger
vein, and training data to get accurate from each test method.
Vocabulary pattern recognition process consists of 3 main
processes: preprocessing aimed at improving the image for
vascular patte
rning, then feature extraction to get feature of blood
vessel pattern, and the last stage is to do similarity between data
testing and training data. So that will get the accuracy of each
method implemented on each dataset.
The data used for the experiment
s are 35 individual data
of palm scan and finger vein. For the average of accuracy test
results of the introduction of palm vein using LLBP feature extraction method is 85.72%. As for the introduction of finger vein using LLBP method is 56.57%. For the ave
rage accuracy on the introduction of palm vein using the SIFT method is 58.93%. The last test that is for the introduction of finger vein using SIFT method has an average accuracy of 80.72%. From the experimental result it can be concluded that LLBP method is more suitable for the introduction of palm vein, while SIFT method is more suitable for finger vein recognition.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Uncontrolled Keywords: | Local Line Binary Pattern (LLBP), Scale Invariant Feature Transform (SIFT), palm vein, finger vein, similarity, matching, operator LLBP, threshold, pembuluh darah. |
| Subjects: | Q Science > QA Mathematics > QA76 Computer software T Technology > T Technology (General) |
| Divisions: | Faculty of Information Technology > Informatics Engineering > 55201-(S1) Undergraduate Thesis |
| Depositing User: | Aini Lusiana Nurul |
| Date Deposited: | 06 Sep 2017 07:34 |
| Last Modified: | 05 Mar 2019 03:58 |
| URI: | http://repository.its.ac.id/id/eprint/42631 |
Actions (login required)
 |
View Item |