Title :
A very fast Kalman filter for image restoration
Author :
Zhang, Jin Yun ; Steenaart, Willem
Author_Institution :
Dept. of Electr. Eng., Ottawa Univ., Ont., Canada
Abstract :
The application of 2-D Kalman filtering to the restoration of images degraded by linear space invariant blur and additive white Gaussian noise is described. R.P. Roesser´s 2-D local state space model (1975) is used to represent the image process and the blur process. As a result, a simple procedure for establishing the Kalman filter equations is obtained. This scalar filtering algorithm provides a computationally feasible procedure for the restoration of large images. To speed up the Kalman filtering procedure, a VLSI systolic array structure is presented. For higher speed and higher utilization of this processor, a diagonal scanning method is suggested. The filter scheme can be easily extended to the causal image model and the causal blur model with nonsymmetric half-plane support
Keywords :
Kalman filters; filtering and prediction theory; picture processing; state-space methods; systolic arrays; white noise; 2-D Kalman filtering; 2-D local state space model; VLSI systolic array structure; additive white Gaussian noise; causal blur model; causal image model; diagonal scanning method; fast Kalman filter; image restoration; linear space invariant blur; nonsymmetric half-plane support; scalar filtering algorithm; Additive white noise; Degradation; Equations; Filtering algorithms; Image restoration; Kalman filters; Nonlinear filters; State-space methods; Systolic arrays; Very large scale integration;
Conference_Titel :
Circuits and Systems, 1990., IEEE International Symposium on
Conference_Location :
New Orleans, LA
DOI :
10.1109/ISCAS.1990.111999
