Robust Visual Tracking with Reliable Object Information and Kalman Filter

. 2021 Jan 28;21(3):889. doi: 10.3390/s21030889

Algorithm 1 Robust Visual Tracking with Reliable Object Information and Kalman Filter

Input: Initial target position

p_{0}

Output: Estimated target position

p_{t}

and updated correlation filters

1:
Initialize the filters according to $p_{0}$ , and save object features $X_{0}$
2:
repeat
3:
According to the $p_{t - 1}$ and correlation filters, calculate the ${\bar{p}}_{t}$ in frame t;
4:
Taking the computed ${\bar{p}}_{t}$ as observation, estimate the target position ${\hat{p}}_{t}$ by Kalman filter;
5:
Fuse the results of two modules, and obtain $p_{t}$ ;
6:
According to the fusion confidence map, analyse the reliability of the tracking process;
7:
if $r e l i a b i l i t y > T h r e s h o l d$ then
8:
Send $X_{0}, X_{t}$ into the online training module, and update the filters;
9:
else
10:
Continue;
11:
end if
12:
until The last frame of the sequences