Putra, Chrystia Aji (2015) Pengenalan karakter plat nomor kendaraan menggunakan extreme learning machine. Masters thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
2211205013-Dissertation.pdf Download (2MB) | Preview |
Abstract
Plat nomor kendaraan adalah salah satu jenis tanda identifikasi kendaraan
bermotor. Pengenalan karakter merupakan salah satu tahap dalam sistem
pengenalan plat nomor. Pengenalan karakter dilakukan untuk mendapatkan data
karakter teks. Metode baru dari Jaringan Syaraf Tiruan(JST) diusulkan untuk
pengenalan karakter pada penelitian ini. Metode yang diusulkan adalah Extreme
Learning Machine (ELM). ELM merupakan jaringan syaraf tiruan feedforward
dengan satu hidden layer. ELM lebih dikenal dengan istilah single hidden layer
feedforward neural network (SLFNs). Metode ELM mempunyai kelebihan dalam
learning speed. Penelitian ini menyelesaikan langkah pengenalan karakter huruf
dan angka pada citra plat nomor. Data karakter didapatkan menggunakan vertical
and horizontal projection. Pengujian pada penelitin ini menggunakan data plat
nomor yang berlaku di Indonesia. Uji coba dilakukan dengan menggunakan 40 citra
plat nomor. Terdapat 273 karakter dalam 40 data plat nomer tersebut. Pengujian
dilakukan pada 273 data karakter. Pengujian dilakukan dengan membandingkan
performa dua buah metode. Dua metode tersebut adalah ELM dan Neural Network
(NN). Pelatihan menggunakan ELM menunjukan waktu pelatihan yang lebih cepat
daripada NN. Pelatihan menggunakan ELM membutuhkan lama waktu 11 detik.
Pelatihan menggunakan NN membutuhkan waktu 144 detik. Akurasi pengujian
ELM dan NN menunjukan hasil yang sama yaitu 78,75%.
==============================================================================================
Vehicle license plate is a type of vehicle identification. Character
recognition is one step in a vehicle license recognition system. Character
recognition is performed to obtain a text character data. Character recognition in
this research proposed a new method of Artificial Neural Network (ANN). The
proposed method was Extreme Learning Machine (ELM). ELM is a feed-forward
neural network with one hidden layer. ELM is better known as single hidden layer
feed-forward neural network (SLFNs). ELM has the advantage in speed learning.
Character data was obtained using vertical horizontal projection. Testing on this
research was done by using vehicle license plates prevailing in Indonesia. Testing
was done by using 40 images of vehicle license plates. There were 273 characters
in 40 images of vehicle license plates. Testing data was conducted on 273
characters. Testing was done by comparing two methods. These methods were
ELM and Neural Network (NN). ELM showed faster training time than that of NN.
Training using ELM took 11 seconds, while training using NN took 144 seconds.
The accuracy of ELM and NN showed the same result, which was 78.75%.
| Item Type: | Thesis (Masters) |
|---|---|
| Additional Information: | RTE 006.42 Put p |
| Uncontrolled Keywords: | Extreme learning machine; pengenalan karakter; plat nomor kendaraan |
| Subjects: | Q Science > Q Science (General) > Q337.5 Pattern recognition systems |
| Divisions: | Faculty of Electrical Technology > Electrical Engineering > 20101-(S2) Master Thesis |
| Depositing User: | - Taufiq Rahmanu |
| Date Deposited: | 28 Oct 2019 03:50 |
| Last Modified: | 28 Oct 2019 03:50 |
| URI: | http://repository.its.ac.id/id/eprint/71444 |
Actions (login required)
 |
View Item |