PELACAKAN BENDA BERGERAK MENGGUNAKAN METODE MEANSHIFT DENGAN PERUBAHAN SKALA DAN ORIENTASI; MOVING OBJECTS TRACKING USING MEAN-SHIFT METHODE WITH SCALE AND ORIENTATION CHANGES

MUHAMMAD IZZUDDIN MAHALI, Agus Harjoko

2013 | Disertasi | PROGRAM STUDI S2 ILMU KOMPUTER

Object tracking is a process to follow the position of objects in an image. Meanshift algorithm is one object tracking algorithm that is often used in the process of tracking an object. Mean-shift algorithm is a non-parametric algorithm is effective and fast but have not been able to follow an object that scale and orientation changes. In this research, the development of methods of classical mean-shift object tracking that is capable of handling the scale and orientation changes. With weights derived image of the target object and the target object candidate to represent the possibility of the region is the target object. Object tracking using mean-shift algorithm uses zero order moments and the first order moment of the weight image. With the zero order moments and the Bhattacharyya coefficient between the target models and candidate models can be used to determine changes in the scale and orientation of the target object. The test results showed a success rate of tracking objects in get some sample videos. From the test results for the system with the input parameter area increment 5 results in successful get is 66,7%, 13,3% partly successful. And 20% failed. As for the input parameter with 10 increment results in successful get is 80%, 13,3% and 6,7% successful partly failed. Level of success in recognizing a target object to be increased when the input parameter area increment will be increased but the addition of time tracking. The larger the area of the target object to be tracked then the time required for tracking is also increasing

Kata Kunci : pelacakan; video; mean-shift; sekala; orientasi