Introduction
Object tracking is an important topic of computer vision. Especially in the fields of surveillance, traffic control, human-robot interaction, and medical imaging. Tracking is the process of estimating the trajectory of an object in subsequent frames, given the bounding box (i.e. position and size) of the object in the initial frame.
Closely related are feature detectors; those will just detect a specific object on each individual frame, based purely on matching features. Pure trackers aim to follow the trajectory as well as predict the future location of the object. This eliminates the risk of mistaking the tracked object for a visually similar object, hereby making the object tracking more stable.
There are many variables affecting the performance of trackers, such as:
- Illumination variations,
- Scale variations,
- Object rotation,
- Occlusion,
- Background clutter, and
- Camera movement.
Currently, there is a wide range of different tracking algorithms, but none can cope with all those problems simultaneously. Due to the importance of the object tracker in numerous applications, an increasing amount of time and effort is…
