Aruco Marker Kalman filter tracking
In this video, the detection of Aruco marker is done is OpenCV, then, the tracking in the image plane is achieved via Kalman filter with a constant-velocity model in Matlab.
Where each corner of the marker (including center) is assigned to a Kalman filter.
With this approach:
#) Cluttering and measurement noise avoided/filtered.
#) In case of no detection, the maker position is predicted correctly
for a short time window.
#) The displacement rate in the image plane can be estimated (see Figure at the end for better insight).
However, this scheme is not perfect and plenty of room available for enhancement, such:
#) In the case of prediction (no measurements), the geometric
constraints of the marker should be taken into account.
#) Choosing a transition model that is dependent on the movement
nature of the target (i.e. Constant velocity model is not suitable for
a rapid movement).