Imron, Muhamad Al (2015) Pendeteksian Dan Pelacakan 2d Multi Objek Secara Real-Time Dengan Menggunakan Kamera Tunggal. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
2109100142-Undergraduate_Thesis.pdf - Accepted Version Download (2MB) | Preview |
Abstract
Dewasa ini teknologi transportasi dunia berkembang
sangat pesat. Salah satu dari teknologi tersebut adalah mendeteksi
dan melacak mobil di sekitar deretan beberapa mobil. Hal tersebut
sangat membantu pengemudi melakukan tindakan tertentu atau
bemanuver untuk menghindari kejadian yang tidak diinginkan
seperti kecelakaan dan bersenggolan. Salah satu metode
pelacakan adalah dengan menggunakan tracking-by-detection,
dimana pelacakan dilakukan pada semua frame yang didapatkan.
Tahapan yang dilakukan dalam proses pendeteksian dan
pelacakan antara lain tahap preprocessing, pendeteksian objek,
dan pelacakan objek. Pendeteksian dan pelacakan dilakukan
berdasarkan bidang kartesian xy yang akan dianalisa pada setiap
frame berdasarkan bentuk dan histogram warna Red Green Blue
(RGB) pada objek.
Dalam penelitian awal ini, objek berbentuk lingkaran
digunakan sebagai penyederhanaan plat nomor yang akan
dideteksi dan dilacak menggunakan kamera tunggal secara realtime.
Multi objek digunakan sebagai pengganti jika kemungkinan
mobil yang melintas lebih dari satu. Hasil pendeteksiaan tersebut
akan diolah dan dianalisa apakah progran tersebut bisa berjalan
sesuai dengan yang diharapkan yaitu tanpa tertukar meski objek
berpapasan.
Hasil yang didapatkan dari pendeteksian secara real-time
tidak ada yang mencapai kemampuan pendeteksian 100%. Dalam
viii
keadaan diam, kemampuan pendeteksian yang diproleh berkisar
diatas 95%. Objek terjauh yang bisa dideteksi secara real-time
yaitu objek dengan radius 20 piksel. Tidak terjadi kesalahan
pendeteksian pada saat objek berpapasan.
===================================================================================================
Nowadays transportation technology is developing very
rapidly. One of these technologies is detection and tracking of a
car around several cars in a street. This technology helps a driver
perform a certain action to avoid unwanted incidents such as
accidents and touched. One of the tracking method is using
tracking-by-detection. This tracking method was done by detecting
all obtained frames in sequence frames . The steps of this process
are the preparation stage, preprocessing, object detection, and
object tracking. The detection and tracking is based on Cartesian
xy, field will be analyzed in each frame based on shape and color
Red Green Blue (RGB) histogram on the object.
In this research, a circle-shaped object was used as a
simplification of the number of object that will be detected and
tracked using a single camera in real-time. Multi object was used
as a substitute if the possibility of a passing object is more than
one. The detection results will be processed and analyzed whether
such programming are able to run as expected.
The results obtained from the detection in real-time have
reached 100% detection capability. In the quiescent state, the
detection capability reach 95%. Most distant object that can be
detected in real-time that is an object with a radius of 20 pixels.
| Item Type: | Thesis (Undergraduate) |
|---|---|
| Additional Information: | RSM 629.892 Ali p |
| Uncontrolled Keywords: | mendeteksi, melacak, detection-by-tracking, frame, warna Red Green Blue(RGB), real-time, multi objek |
| Subjects: | T Technology > TJ Mechanical engineering and machinery |
| Divisions: | Faculty of Industrial Technology > Mechanical Engineering > 21201-(S1) Undergraduate Thesis |
| Depositing User: | Yeni Anita Gonti |
| Date Deposited: | 13 Feb 2020 02:37 |
| Last Modified: | 13 Feb 2020 02:37 |
| URI: | http://repository.its.ac.id/id/eprint/74861 |
Actions (login required)
 |
View Item |