Within our project we were supposed to trace two objects (cars) that facing contractions from their surroundings and by each other.
To trace those cars, we used some mathematical models that help with identifying object and estimating their position when not found.
To find the object we used cross correlation method, that searched for the most similar window in the frame in compare to our object, once we found a window that returned a high cross correlation factor, we easily tracked the object.
On the other hand, when we couldn’t find the object because of contractions or sharp change, we started estimating their position use a fixed velocity model that was built by previous samples of the objects.
We also used some methods that helped us to reduce the searching area within the frame to accelerate the searching process.