Yudha, Ery Permana (2018) Sistem Temu Kembali Citra Berdasarkan Kemiripan Citra Terhadap Kelas Menggunakan Ekstraksi Fitur SURF. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
05111340000117-Undergraduate_Thesis.pdf - Accepted Version Download (2MB) | Preview |
Abstract
Temu kembali citra merupakan proses mencari dan menampilkan citra yang relevan terhadap citra query dari
sekumpulan citra dalam database. Dalam mendapatkan fitur ada banyak cara seperti warna, tekstur, dan bentuk. Sedangkan cara yang sering digunakan dalam temu kembali citra yaitu dengan mengurutkan citra berdasarkan kemiripan citra terhadap citra (image-to-image) seperti kemiripan antara citra query dengan citra database. Cara ini mengabaikan informasi inheren yang ada pada kelas pada citra query (query class). Maka dari itu, pada tugas akhir ini diajukan metode temu kembali citra berdasarkan kemiripan citra terhadap kelas (imageto-class). Metode ini membutuhkan fitur yang dihasilkan dari ekstraksi fitur menggunakan Speeded Up Robust Features (SURF) berupa interest points. Setelah itu, fitur tersebut digunakan pada proses image-to-class menggunakan metode RICS. Metode ini menggunakan kerangka kerja pengulangan (iterative framework) di mana ukuran dari query class makin bertambah berdasarkan hasil temu kembali sebelumnya dan daftar peringkat dihasilkan berdasarkan kemiripan diantara citra pada database dengan query class. Metode ini mampu mengetahui informasi inheren pada query class, maka dari itu dapat membantu meningkatkan hasil temu kembali yang akurat. Hasil dari metode ini menghasilkan nilai precision 86,00% dan nilai recall 86,00%.
=====================================================================================================
Image retrieval is the process of finding and displaying images that match the image query of image database. There are many ways to get features like colors, texture, and shapes. Meanwhile, the way that usually used in image retrieval is to rank the images according image to image similarity such as the similarities between query image and images in the database. This way ignores the inherent information of the class that query the query image belongs to (query class). Therefore, the retrieval method based on image to class similarity is presented in this undergraduate thesis. This method need interest points features generated from feature extraction using Speeded Up Robust Features (SURF). Afterwards, the features are used for image to class process using RICS algortihm. This method uses an iterative framework, where the size of the query class is enlarged according to the previous retrieval results and the ranked list is generated according to the similarities between the image in the database and the query class. This method enables to explore the inherent information of the query class and helps to improve the retrieval accuracy. The result of this method is 86,00% of precision and 86,00% of recall.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | RSIf 005.74 Yud s-1 3100018076563 |
| Uncontrolled Keywords: | temu kembali citra, query class, image-to-class, SURF, RICS |
| Subjects: | T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques |
| Divisions: | Faculty of Information and Communication Technology > Informatics > 55201-(S1) Undergraduate Thesis |
| Depositing User: | Ery Permana Yudha |
| Date Deposited: | 17 Oct 2020 03:00 |
| Last Modified: | 17 Oct 2020 03:00 |
| URI: | http://repository.its.ac.id/id/eprint/55775 |
Actions (login required)
 |
View Item |