A New Kalman Filter Based 2D AR Model Parameter Estimation Method

This paper presents a new method based on the Kalman filter (KF) for the two-dimensional (2D) autoregressive model parameters estimation. For this purpose, the corresponding relations of KF in one-dimensional (1D) case are extended to the 2D case and the related algorithm is presented. Main advantages of the proposed method are: online and recursive parameter estimation, and detection of parameters changes. Meantime, because the presented approach does not involve complex and time consuming matrix calculations, so it is computationally efficient. Numerical examples are expressed to verify the efficiency of the proposed method.