A multiple point correlation tracker for time-to-go estimation

It is shown how it is possible to estimate time-to-go to a landmark using the output from a multiple point correlation (MPC) tracker. It is shown how the correlation process is equivalent to a classical parameter estimation problem where the horizontal and vertical shifts between reference and search images are the state variables. A model of the motions between frames is postulated and analytical expressions derived for the expected value of the time-to-go. These expressions are shown to agree well with the results of simulation. Synthetic image sequences are used to test the MPC algorithm and accurate estimates of time-to-go are obtained.<
>