Object Tracker-A Real Time Object Tracking System with Particle Filter

The filtering problem is to estimate the motion of an object detected by sensors, cameras, voice recorders, etc., based on its specific characteristics such as color, light reflection, and sound waves. There are many filtering algorithms that have been used by researchers in the past to solve the related problems. In the literature, Lucas Rod’es Guirao (partner Diego Mor’ın) conducted an experiment to estimate the motion of an object and compared the performance of the Kalman filter and the particle filter in their study. They conclude that using the Kalman filter after the particle filter can slightly improve the performance of the particle filter alone, but the combination of Kalman filter and particle filter is still very sensitive to particle deprivation when there are occlusions. The objective of this thesis is to use the particle filter algorithm to improve the tracking mechanizm in the way that updating particles which tracking the movements of the ball in a faster way. This thesis presents the modified particle filter optimization for tracking the object efficiently and follow its posterior movements on the exact point to track the object truly. Depending on the proposed particle filter algorithm, the camera is set to track the object behind of the conjectural goal line and the proposed code performs the estimation of particles as a guider of goalkeeper. Focused small ball considered as an object to estimate its x and y coordination in a two-dimensional area captured by a fixed camera connected to the computer. A 30- second live shot is used to track the ball by using the RGB features of the particles to calculate the estimation of the object by a coded Matlab application. The proposed agorithm checks whether the ball has crossed the goal line or not, and then print these all output information on terminal section of Matlab.