Al-Habib, Hasanuddin (2015) Super Resolusi Objek Berbasis Citra Tracking Menggunakan Metode Phased Based Image Matching Dan Metode Proyeksi Pada Himpunan Convex. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
1211100083-Undergraduate-Thesis.pdf Download (3MB) | Preview |
Abstract
Super resolusi merupakan teknik untuk mendapatkan citra beresolusi
tinggi dari citra yang beresolusi rendah. Citra resolusi rendah yang
digunakan dapat berupa citra tunggal maupun rangkaian citra yang
diambil dari scene yang sama agar citra tersebut menyediakan informasi
yang sama untuk proses rekontruksi citra resolusi tinggi. Pada umumnya
super resolusi dilakukan pada keseluruhan piksel frame, padahal
kebutuhan informasi dari suatu citra hanya terdapat pada suatu bagian
tertentu dari frame tersebut bukan pada keseluruhan piksel frame. Oleh
karena itu, agar proses super resolusi dapat lebih efektif maka digunakan
ROI (Region Of Interest) untuk mendapatkan bagian dari frame dan
melakukan proses tracking dalam rangkaian citra. Pada tugas akhir ini
digunakan metode phased based image matching untuk proses registrasi
dan metode projection onto convex sets untuk proses rekonstruksi. Pada
pengujian yang dilakukan nilai rata-rata PSNR diperoleh 30,3183 dB
dan nilai PSNR akan semakin besar jika citra observasi semakin banyak.
Selain itu, diperoleh waktu komputasi citra tracking sebesar 7,456 detik
sedangkan pada keseluruhan piksel pada frame citra sebesar 188,306
detik. Hal ini menunjukkan bahwa waktu komputasi citra tracking lebih
efektif dibandingkan keseluruhan piksel pada frame citra
==================================================================================================
Super-resolution is a technique to obtain high-resolution
images from low-resolution image. Low-resolution images used
may be a single image or series of images taken of the same scene
so that the image provides the same information for a high
resolution image reconstruction process. In general, carried out
on the overall super-resolution pixel frame, whereas the
information needs of an image only in a certain part of the frame
rather than the whole pixel frame. Therefore, in order to superresolution
process can be more effective then used ROI (Region
Of Interest) to get a piece of the frame and make the process of
tracking in a series of images. In this final project, its
implemented phased based image matching method for the
registration process and the projection onto convex sets method
for the reconstruction process. In the tests performed, we got the
average of PSNR value is 30,3183 dB and this PSNR value will
increase if the number of observations image increase. In
addition, images obtained by computing time tracking of 7.456
second while the total pixels in the image frame at 188.306
second. This indicates that the computing time tracking image is
more effective than a whole pixel in the image frame.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | RSMa 511.322 Alh s |
| Uncontrolled Keywords: | Objek tracking, Super resolusi, normalized cross correlation, phased based image matching, projection onto convex sets, ROI. |
| Subjects: | T Technology > TR Photography > TR267.733.M85 Multispectral imaging |
| Divisions: | Faculty of Mathematics and Science > Mathematics > 44201-(S1) Undergraduate Thesis |
| Depositing User: | Mr. Marsudiyana - |
| Date Deposited: | 05 Dec 2019 04:06 |
| Last Modified: | 05 Dec 2019 04:08 |
| URI: | http://repository.its.ac.id/id/eprint/72212 |
Actions (login required)
 |
View Item |