Abstract
In this paper, we propose an estimation method of system parameters for a two-dimensional noisy image and apply the estimation results to restoration of the noisy image data. The unknown parameters required for orthogonal transform are first roughly estimated by the Yule-Walker equation. Using these estimated parameters, the noisy image data are transformed into frequency domain by FFT. The system parameters of the two-dimensional image are estimated by using adaptive digital filter (ADF) in frequency domain. The noisy image in frequency domain is restored by a filtering algorithm based on the system parameters and innovation process obtained by the ADF. The restored image in frequency domain is transformed into spatial domain by inverse FFT. Simulation results show the effectiveness of the present algorithm in comparison with conventional methods.
