Robust modeling edge adaptive reduced update Kalman filter

Abstract

The proposed image reconstruction algorithm is a combination of the robust modeling modified reduced update Kalman filter (RUKF) introduced by Belaifa and Schwartz (1992) and the edge adaptive RUKF by Tekalp et al. (1989), the filter is designed to remove white Gaussian noise as well as salt and pepper noise from a corrupted image while reducing the edge smoothing effect associated with the RUKF. Using the edge adaptive RUKF, the proposed algorithm selects the model that best describes the pixel from the five Kalman filter models available. It uses Kalman edge models when filtering edges and the robust model when filtering non-edge pixels.