Sukmawati, Ayu Kardina (2017) Implementasi Deteksi Seam Carving Berdasarkan Perubahan Ukuran Citra Menggunakan Local Binary Patterns dan Support Vector Machine. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
5113100072-Undergraduate_Theses.pdf - Published Version Download (3MB) | Preview |
Abstract
Seam carving adalah metode yang digunakan untuk content-aware image resizing. Seam carving bertujuan untuk mengubah ukuran citra atau image resizing dengan tidak menghilangkan konten penting yang ada pada citra. Dalam bidang forensik digital, seam carving banyak dibahas khususnya tentang deteksi seam carving pada citra. Hal tersebut bertujuan untuk mengetahui apakah suatu citra sudah pernah melalui proses pengubahan ukuran menggunakan seam carving atau belum.
Tugas akhir ini mengusulkan sebuah metode deteksi seam carving berdasarkan perubahan ukuran citra menggunakan Local Binary Patterns dan Support Vector Machine. Citra yang akan dideteksi dihitung variasi teksturnya menggunakan Local Binary Patterns. Proses selanjutnya adalah ekstraksi fitur dari distribusi energy yang menghasilkan 24 fitur. Data fitur citra selanjutnya dilakukan proses normalisasi. Uji coba fitur menggunakan k-fold cross validation dengan membagi data menjadi training dan testing. Selanjutnya data tersebut akan memasuki proses klasifikasi menggunakan Support Vector Machine dengan kernel Radial Basis Function.
Uji coba dilakukan terhadap citra asli dan citra seam carving. Citra seam carving yang digunakan dibedakan
viii
berdasarkan skala rasionya yaitu 10%, 20%, 30%, 40%, dan 50%. Jumlah data yang digunakan adalah sebanyak 400 citra untuk setiap uji coba pada tiap skala rasio dengan menggunakan 10-fold cross validation. Rata-rata akurasi terbaik yang dihasilkan sebesar 73,95%.
========================================================================================================================
Seam carving is method used for content-aware image resizing. Seam carving is designed to resize the image by not eliminating the important content that is in the image. In digital forensic area, seam carving is much discussed, especially about seam carving detection in image. It aims to determine whether an image has been through the process of resizing using seam carving or not.
This final project propose a method of seam carving detection based on image resizing using Local Binary Patterns and Support Vector Machine. The texture variation of image will be calculated using Local Binary Patterns. The next process is extraction process from energy distribution and produces 24 features. The feature data is then performed normalization process. The normalized features will be tested using k-fold cross validation by dividing the data into training and testing. Finally, the data will be classified using Support Vector Machine with Radial Basis Function kernel.
The trial test use original image and seam carved image. Seam carved image used is differentiated by its scaling ratios of 10%, 20%, 30%, 40%, and 50%. There are 400 image used for each trial test on each scaling ratios by using 10-fold cross validation. The best average accuracy is 73,95%.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Uncontrolled Keywords: | seam carving, Local Binary Patterns, k-fold cross validation, Support Vector Machine, Radial Basis Function kernel |
| Subjects: | T Technology > T Technology (General) Z Bibliography. Library Science. Information Resources > ZA Information resources > ZA4450 Databases |
| Divisions: | Faculty of Information Technology > Informatics Engineering > 55201-(S1) Undergraduate Thesis |
| Depositing User: | Ayu Kardina Sukmawati |
| Date Deposited: | 18 Aug 2017 06:59 |
| Last Modified: | 05 Mar 2019 03:28 |
| URI: | http://repository.its.ac.id/id/eprint/42339 |
Actions (login required)
 |
View Item |