A new approach to 2-D Kalman filtering
Jiang Min, Xiang Chen · 2005
A new approach is made to 2-D Kalman filtering, which consists of three parts sequentially. First, we simplify the model of the reduced update kalman filter (RUKF) further. Then, we present a fast recursive least-sequare identification (FRLSI) to estimate paramenters of space-variant images on-line. Finally, we design a genearalized likelihood ratio (GLR) edge detector to eliminate the edge smear. Experiment shows that our improvements are feasible in practice and give better results.