PENGENALAN TULISAN AKSARA SUNDA OFFLINE DENGAN JARINGAN SYARAF TIRUAN BACKPROPAGATION

    Dea rokhmatun Iradewa, - (2014) PENGENALAN TULISAN AKSARA SUNDA OFFLINE DENGAN JARINGAN SYARAF TIRUAN BACKPROPAGATION. S1 thesis, Universitas Pendidikan Indonesia.

    Abstract

    Pengenalan tulisan tangan secara offline adalah sebuah teknik dimana input data berupa gambar karakter tulisan hasil akuisisi menggunakan scanner atau kamera digital yang dikenali komputer sebagai sebuah himpunan piksel. Kompleksitas pada proses pengenalan tulisan tangan semakin bertambah menginggat masing-masing sistem keaksaraan memiliki kaidah penulisan tersendiri ditambah dengan variasi tulisan dari setiap penulis yang memiliki karakteristik tersendiri. Dalam penelitian ini dilakukan pengenalan terhadap tulisan tangan aksara sunda secara offline dengan menggunakan metode Image Centroid and Zone (ICZ) dan Zone Centroid and Zone (ZCZ) dalam proses ekstraksi fitur dan dilanjutkan dengan menerapkan metode Jaringan Syaraf Tiruan (JST) Backpropagation dalam proses klasifikasi. Hasil pengujian dengan data yang sama dengan data pelatihan menghasilkan akurasi pengenalan rata-rata sebesar 99,89 %. Sedangkan pada pengujian dengan menggunakan k-fold validation dengan nilai k=6 menghasilkan akurasi pengenalan rata-rata sebesar 91,89 %.

    ;---Offline handwriting recognition is a technique which the input data is image of handwriten character image that acquisition using scanner or digital camera. That image recognized by the computer as a set of pixels. The complexity of the process of handwriting recognition is become more complex considering each literacy system has its own rules of writing and every writer has its own characteristics and variations while writting. In this research conducted offline handwritten character recognition of Sundanese script using Image Centroid and Zone (ICZ) and Zone Centroid and Zone (ZCZ) in the process of feature extraction and followed by applying the method of Artificial Neural Network (ANN) backpropagation in the classification process. The Test results with the same data as the training data generating an average recognition accuracy of 99.89%. While on testing using the k-fold validation with k = 6 shows average recognition accuracy of 91.89%.

    Baca Full Text klik disini
    [thumbnail of S_KOM_0700940_Title.pdf] Text
    S_KOM_0700940_Title.pdf
    Download (320kB)
    [thumbnail of S_KOM_0700940_Abstract.pdf] Text
    S_KOM_0700940_Abstract.pdf
    Download (124kB)
    [thumbnail of S_KOM_0700940_Table_of_content.pdf] Text
    S_KOM_0700940_Table_of_content.pdf
    Download (235kB)
    [thumbnail of S_KOM_0700940_Chapter1.pdf] Text
    S_KOM_0700940_Chapter1.pdf
    Download (145kB)
    [thumbnail of S_KOM_0700940_Chapter2.pdf] Text
    S_KOM_0700940_Chapter2.pdf
    Restricted to Staf Perpustakaan
    Download (879kB)
    [thumbnail of S_KOM_0700940_Chapter3.pdf] Text
    S_KOM_0700940_Chapter3.pdf
    Download (279kB)
    [thumbnail of S_KOM_0700940_Chapter4.pdf] Text
    S_KOM_0700940_Chapter4.pdf
    Restricted to Staf Perpustakaan
    Download (1MB)
    [thumbnail of S_KOM_0700940_Chapter5.pdf] Text
    S_KOM_0700940_Chapter5.pdf
    Download (58kB)
    [thumbnail of S_KOM_0700940_Bibliography.pdf] Text
    S_KOM_0700940_Bibliography.pdf
    Download (64kB)
    Official URL: http://repository.upi.edu
    Item Type: Thesis (S1)
    Additional Information: No. Panggil : S KOM DEA p-2014; Pembimbing : I. Wawan Setiawan, II. Eddy Nugroho; NIM : 0700940.
    Uncontrolled Keywords: aksara sunda, backpropagation, pengenalan tulisan, ICZ, ZCZ, sundanese script, backpropagation, character recognition, ICZ, ZCZ.
    Subjects: T Technology > T Technology (General)
    T Technology > TP Chemical technology
    Divisions: Fakultas Pendidikan Matematika dan Ilmu Pengetahuan Alam > Program Studi Ilmu Komputer
    Depositing User: YUSUP REZA ADITYA HARDIAN
    Date Deposited: 13 Feb 2020 07:47
    Last Modified: 13 Feb 2020 07:47
    URI: http://repository.upi.edu/id/eprint/44845

    Actions (login required)

    View Item View Item