Aiming to the slow tracking and steadystate errors that exist when UKF(unscented Kalman filter) applied in nonlinear systems state estimation, an algorithm of video target tracking using strong tracking UKF was presented. Based on the unscented transformation, combined with the advantages of strong tracking filter and UKF filter, this algorithm adjusts Kalman gain with introducing timevarying fading factor in priori error covariance matrix to force the output residuals to maintain orthogonal, extract useful information in the residuals and improve the tracking capability to status change. The simulation results show that, moving target tracking in vedio utilizing strong tracking UKF has higher tracking accuracy and smaller MSE in state filtering.