Developing a Real-Time Object Tracking System

Abstract

Object detection and tracking is an important task within the field of computer vision, due to its promising applications in many areas, such as video surveillance, human-computer interaction, intelligent transportation, etc. The availability of high quality and inexpensive video cameras and the increasing need for automated video analysis has generated a great deal of interest in the areas of motion tracking. The commonly used technique to extract the foreground objects is through subtracting the background. Therefore, the basic operation needed is the separation of the moving objects from the background static information. The main goal of the object tracking is to estimate the locations and the motion parameters of the target, in an image sequence, given the initial position in the first frame. Most of the current techniques modeled complex frameworks, with more accurate results, which are not suitable for real-time applications. On the other hand, simple detection and tracking techniques were proposed with low accurate results. There are many sources of uncertainty for the object locations, such as illumination variation, measurement noise, moving object’s shadow, addition or removal of stationary objects and scene motion. There exists no single detection or tracking technique that can be successfully applied to all tasks and situations.