PENGENALAN KARAKTER TULISAN TANGAN AKSARA LAMPUNG MENGGUNAKAN EKSTRAKSI FITUR CHAIN CODE 8 DIRECTION DAN PROJECTION PROFILE

TRI LESTARI, 1417051143 (2018) PENGENALAN KARAKTER TULISAN TANGAN AKSARA LAMPUNG MENGGUNAKAN EKSTRAKSI FITUR CHAIN CODE 8 DIRECTION DAN PROJECTION PROFILE. FAKULTAS MATEMATIKA DAN ILMU PENGETAHUAN ALAM, UNIVERSITAS LAMPUNG.

[img]
Preview
Text
ABSTRAK.pdf

Download (56Kb) | Preview
[img] Text
SKRIPSI FULL.pdf
Restricted to Registered users only

Download (2024Kb)
[img]
Preview
Text
SKRIPSI TANPA BAB PEMBAHASAN.pdf

Download (1379Kb) | Preview

Abstrak

perkembangan zaman, pengetahuan akan Aksara Lampung semakin tergerus. Salah satu upaya untuk melestarikan aksara tersebut adalah dengan melakukan penelitian tentang pengenalan karakter Aksara Lampung. Penelitian ini merupakan lanjutan dari penelitian sebelumnya tentang pengenalan karakter tulisan tangan Aksara Lampung. Pengembangan dilakukan dari segi penggunaan fitur untuk pengenalan karakter aksara. Fitur-fitur yang digunakan adalah chain code 8 direction, projection profile dan gabungan keduanya. Jumlah karakter pada dataset secara keseluruhan sebanyak 32140 file dengan format grayscale image. Dataset tersebut terbagi menjadi tiga kelompok data yaitu training, validation, dan testing. Sebagai alat klasifikasi digunakan LibSVM (Library Support Vector Machine). Akurasi yang diperoleh pada penggunaan fitur chain code 8 direction sebesar 90,08%, sedangkan penggunaan fitur projection profile menghasilkan akurasi sebesar 79,28%. Penggabungan kedua fitur memberikan hasil akurasi tertinggi dengan peningkatan 4% dari penggunaan fitur chain code 8 direction menjadi 94,00%. Kata kunci : aksara Lampung, chain code 8 direction, fitur, pengenalan karakter tulisan tangan, projection profile. RECOGNITION OF HANDWRITTEN CHARACTER LAMPUNG SCRIPT BY USING CHAIN CODE 8 DIRECTION AND PROJECTION PROFILE FEATURE EXTRACTION Lampung script is one of the cultural heritages in the form of writing character owned by Lampung society. The basic character consists of 20 characters with various shapes. As the times progressed, the knowledge of the Lampung script increasingly faded out. One of the efforts to preserve the script is by conducting research on the recognition of Lampung script characters. This research is a continuation of previous research on the recognition of handwritten characters of Lampung script. This research is evolved from the previous research in terms of the use of features for character recognition. The features used are chain code 8 direction, projection profile and the combination of both. The number of characters in the dataset is 32140 files with grayscale image format. The dataset is divided into three groups of data that are training, validation, and testing. As classification tool, this research used LibSVM (Library Support Vector Machine). The accuracy obtained on the use of chain code 8 direction feature reached 90.08%, while the use of projection profile feature reached the accuracy of 79.28%. Merging the two features results the highest accuracy with 4% increasing number from the use of chain code 8 direction feature becoming 94.00%. Keywords: chain code 8 direction, feature, handwriten character recognition, Lampung script, projection profile.

Tipe Karya Ilmiah: Skripsi
Subyek: Q Science (General)
Q Science (General) > QA Mathematics > QA76 Computer software
Program Studi: Fakultas MIPA > Prodi Ilmu Komputer
Depositing User: 201834131 . Digilib
Date Deposited: 04 Aug 2018 03:44
Last Modified: 04 Aug 2018 03:44
URI: http://digilib.unila.ac.id/id/eprint/32592

Actions (login required)

View Item View Item