Shift variant image degradation and restoration using SVDInternational Journal of Computer Graphics and Image Processing (1979)
Restoration of images degraded by a linear shift variant (LSV) point spread function has been considered. Since the well-developed techniques of Fourier analysis are not directly applicable a numerical solution of the Fredholm integral equation describing the degradation has been attempted. The integral equation is first digitized to get a set of algebraic equations. The degradation matrix is factored using singular value decomposition (SVD). Presence of errors in the degraded image (due to detector noise) sometimes makes the solution obtained by inversion of the degrading matrix almost impossible. A major contribution to the noise in the solution thus obtained is due to the components of the error vector along the coordinate axes corresponding to smaller singular values. The effect of these errors has been minimized by estimating the copponents of the image along these coordinate axes from known statistical properties of images.
Publication DateMarch, 1979
Citation InformationArun D Kulkarni and S. C. Sahasrabudhe. "Shift variant image degradation and restoration using SVD" International Journal of Computer Graphics and Image Processing Vol. 9 Iss. 3 (1979) p. 203 - 212
Available at: http://works.bepress.com/arun-kulkarni/42/