Algorithm 1: Time interval determination

Choose the threshold value s0 experimentally. Choose arbitrary an interval of confidence M0T, where M0 is the arbitrary chosen number of frames. Waiting when the signal $x_{i,j}\left(nT\right)$/$y_{i,j}\left(nT\right)$ is below the threshold value s0. If the signal value $x_{i,j}\left(nT\right)/y_{i,j}\left(nT\right)$ exceeds the threshold value s0 during the confidence interval [NbT, NbT+M0T] then moment NbT-M0T determines beginning of signal pattern. If the signal value $x_{i,j}\left(nT\right)$/$y_{i,j}\left(nT\right)$ is below the threshold value s0 during the interval of confidence [NeT, NeT+M0T] then moment NeT+M0T is chosen as the end of the signal pattern. The cross correlation function is calculated as the following: $$R_{i,j}\left(kT\right)=\frac{1}{N_e-N_b+2M_0}{\displaystyle \sum }_{n=N_b-M_0}^{N_e+M_0}{x}_{i,j}\left(nT\right)y_{i,j}\left(\left(n-k\right)T\right),k=\dots ,-1,0,1\dots$$ Finally, velocity of the flow in (i,j) pixel is calculated according to Equation (2), however adding the interval [NeT, NeT+M0T]: $$\overline{V_{i,j|\left[N_{e}T,N_{e}T+M_{0}T\right]}}=\raisebox{1ex}{$d$}\!\left/ \!\raisebox{-1ex}{${\tau }_0$}\right.$$ Return to step 3 where n=Ne i.e., starting from the moment of the interval when the signal $x_{i,j}\left(N_{e}T\right)$/$y_{i,j}\left(N_{e}T\right)$. is below the threshold value s0.