IMAGE RESTORATION WITH KALMAN FILTERING TECHNIQUES

Abstract

In 1972, Nahi[l 7] first used Kalman filtering technique in image restoration. Since then, many of Kalman filtering algorithms for image restoration have been developed. In this thesis, two new algorithms are developed. The first one, Kalman Smoother (KS), is developed in an effort to improve the quality of the restored image by giving the optimum estimation of a pixel value based on the observation data on "both sides of the pixel" being processed. The second algorithm is Simplified Reduced Order Model Kalman Filter (SROMKF). The SROMKF is based on the Reduced Order Image Model[21]. It can greatly reduce computation load and memory requirement in the filtering process. Our experimental results show that the two new algorithms (Kalman Smoother and Simplified Reduce Order Kalman Filter) are effective and their performances are comparable with other Kalman Filtering algorithms.