In control theory, a state observer is employed to estimate the internal states of the given process. In this paper, a full state observer is engineered to perform predictive video target tracking enhanced with a lowpass filter designed to achieve better tracking quality. The impact of pole placement in the observer is analyzed and the optimal pole values are identified for this particular research. The tests are conducted on two sets of given video sequences, in which a jumping ball and a randomly moving target is respectively appear. The results suggest that the proposed approach boasts technical merits such as low computational complexity, good tracking robustness for highly mobile moving targets, and system noise worry-free.