An algorithm is developed to estimate the original image intensity from a degraded image. The degradation phenomena is a Gaussian noise contaminated by an outlier sequence. The proposed algorithm is a combination of the robust algorithm proposed by Kashyap and Eom and the reduced update Kalman filter (RUKF) developed by Woods and Radewan. The proposed algorithm is compared to some commonly used techniques such as the median filter, the robust algorithm, and the RUKF.
IEEE Transactions on Signal Processing
Department of Systems and Computer Engineering

Belaifa, H.B.H. (Hosni B. H.), & Schwartz, H.M. (1992). Robust Modeling for Image Restoration Using a Modified Reduced Update Kalman Filter. IEEE Transactions on Signal Processing, 40(10), 2584–2588. doi:10.1109/78.157298