Moving object tracking is a method used
to detect and analyze changes that occur in an object from a frame to the next
one in a video 1. The aim of moving object tracking is to estimate the
trajectory of an object in the image plane as it moves around a scene. Moving
object tracking may help to detect, identify, and analyze the object 6.
In the military field, automatic
surveillance video tracking is used to monitor a certain location. For
instance, it can be used either to guard the state boundary or to monitor the
enemy’s movement on the battlefield. CCTV also works in the same manner in
monitoring the crowds such as campus, mall, market, etc. In soccer video,
object tracking is used to know a player real-time position. Thus, a sports
analyst will be able to inform the game pattern, the formation, and the
strategy that should be done. Overall, the game should involve moving object
tracking. The observed object is relatively small compared to the relatively
big video size.
The methods or approaches used to track
the moving object are SIFT, Mean-Shift, Camshift, Kalman Filter, and Particle
Filter. Each method has their own advantages and disadvantages. For example,
Kalman Filter can only be used in a linear or stable system and with noise
distributed normally (Gaussian) 2. The Camshift method is a development of
Mean-shift. Camshift can be used for the general purpose of moving object
tracking 4. However, this method cannot detect an object hindered by another
object within a frame. Otherwise, Kalman Filter is able to predict such an
object. Thus, a combination of Camshift and Kalman Filter are expected to ease
the moving object tracking in various situation and condition 3.
A research on moving object tracking
using Kalman Filter had been conducted with the title ‘Moving Object Tracking
Using Kalman Filter 5. This work was done on an object having various
background targeting a single object. Another work titled ‘Moving Target
Tracking Based on Camshift Approach and Kalman Filter 3 succeeded in
combining Camshift and Kalman Filter. In this work, Kalman Filter is used as
the tracking method and fixed by the Camshift method.
This paper discuss the results of moving single
object tracking in a video with Camshift method as the main tracking and Kalman
filter for prediction and correction. Obstacles were found in the video
including objects moving together, the similar color of the object and the
background, and moving camera. Video acquisition process obtained in State
Islamic Institute of Tulungagung and video dataset 6 from a valid source.