On the possibilities of online image restoration by means of a finite impulse response filter
Abstract
In order to restore blurred images in a video system online, the design of (digital) twodimensional, finite impulse response filters was investigated. With respect to the computation time, the maximum extension is limited to 9 by 9 elements. Linear restoration filters normally possess an infinite impulse response. In the case of a spatially invariant, circular symmetric blur function, the frequency transfer function (ftf) of an infinite impulse response filter was approximated in a weighted mse sense by the ftf of a finite impulse response filter. A program was developed to compute these finite impulse responses. The truncation of the impulse response elements is performed by a special designed algorithm to keep the truncation error small. From the results of the approximations it can be concluded that in general restoration filters can be approximated well, in some cases even for a smaller number of impulse response elements then 9 by 9. In an experiment with subjects, which restoration filters perform optimal restorations was investigated with respect to the subjective image quality.
 Publication:

Ph.D. Thesis
 Pub Date:
 August 1977
 Bibcode:
 1977PhDT........45S
 Keywords:

 Image Enhancement;
 Impulses;
 Responses;
 Restoration;
 Blurring;
 Computer Programming;
 Digital Filters;
 Image Filters;
 Images;
 Optimization;
 Real Time Operation;
 Transfer Functions;
 Optics