Kalman filter formulation of low-Level television image motion estimation

The problem of television image motion estimation is formulated as an application of Kalman filter theory. The nonuniform image motion present in a television scene is represented as the state variable of a randomly driven difference equation. A new approach is then introduced for linearizing measurement equations that arise in low-level image velocity estimation. Kalman filter theory is applied to the problem of optimally solving the nonuniform motion estimation problem based upon the image motion model and the linearized measurement equations.