Mulyasari, Ni Luh Made Asri (2017) Implementasi Deteksi Copy-move Forgery pada Citra Menggunakan Metode Histogram of Oriented Gradients (HOG). Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
5113100085-Undergraduate-Theses.pdf - Published Version Download (2MB) | Preview |
Abstract
Pemalsuan citra terdiri dari bermacam-macam jenis yang salah satunya adalah melakukan duplikasi suatu bagian gambar kemudian ditempelkan pada bagian lain di gambar tersebut. Hal ini disebut sebagai copy-move forgery atau pemalsuan pada citra. Karena bagian yang diduplikasi berasal dari citra yang sama, maka property-properti seperti noise, warna dan tekstur akan kompatibel dengan seluruh citra dan dengan demikian menjadi lebih sulit untuk dibedakan dan deteksi secara kasat mata.
Tugas akhir ini mengusulkan sebuah sistem deteksi pemalsuan copy-move pada citra dengan menggunakan metode Histogram of Oriented Gradients (HOG). Data citra pertama dibagi ke dalam blok-blok yang saling overlap kemudian setiap blok di ekstraksi dengan HOG yang menghasilkan fitur vektor. Kemudian deteksi copy-move dicari dengan menghitung jarak antar fitur setiap blok, blok-blok yang memiliki nilai jarak nol merupakan daerah yang mengalami copy-move.
Uji coba dilakukan terhadap 20 citra yang diambil dari CoMoFod database. Berdasarkan hasil uji coba menunjukkan bahwa metode ini dapat memberikan hasil deteksi copy-move yang akurat dengan rata-rata akurasi, specificity dan sensitivity adalah 99,59%, 99,91%, dan 90,21%.
=============================================================================================
Images forgery are composed of various types, one of which is duplicating a portion of the image to another part of the image. This is called copy-move forgery in the image. Because the duplicated part comes from the same image, properties such as noise, color and texture will be compatible with all images and thus become more difficult to distinguish and perceive. The purpose of copy-move on the image can vary. General is generally done for the addition of objects or to an object in an image.
This final project proposes a copy-move counterfeiting detection system on the image using the Histogram of Oriented Gradients (HOG) method. The first image data is divided into overlaping blocks then each block is extracted with HOG which produces the vector feature. Then copy-move detection is searched by calculating the distance between features of each block, blocks that have a zero distance value are areas experiencing copy-move.
Testing of 20 images from the CoMoFod database. Based on the results shows that this method can provide accurate copy-move detection with average accuracy, specificity and sensitivity is 99.59%, 99.91%, and 90.21%.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Uncontrolled Keywords: | Copy-move, Euclidean Distance, Histogram of Oriented Gradients, image forgery, pemalsuan citra |
| 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: | Ni Luh Made Asri Mulyasari |
| Date Deposited: | 11 Aug 2017 04:33 |
| Last Modified: | 05 Mar 2019 06:31 |
| URI: | http://repository.its.ac.id/id/eprint/42921 |
Actions (login required)
 |
View Item |