Title :
Sensitivity of color LMMSE restoration of images to the spectral estimate
Author :
Trussel, H.J. ; Sezan, M. Ibrahim ; Tran, Duong
Author_Institution :
Dept. of Electr. & Comput. Eng., North Carolina State Univ., Raleigh, NC, USA
fDate :
1/1/1991 12:00:00 AM
Abstract :
The sensitivity to the spectral estimate of the original image of linear minimum mean-square error (LMMSE) color image restoration methods is determined. It is concluded that (i) sensitivity of the full-correlation Wiener restoration to the spectral estimate is higher than that of the independent-channel Wiener restoration, and the full-correlation restoration is extremely sensitive to windowing, (ii) among independent-channel restorations, the ones based on autoregressive spectral estimates are the least sensitive, and (iii) full-correlation Wiener restoration may be outperformed by the independent-channel restoration depending on the prototype image used in spectral estimation
Keywords :
correlation methods; parameter estimation; picture processing; sensitivity analysis; Wiener restoration; colour image restoration; full-correlation restoration; independent-channel restoration; linear minimum mean-square error; multichannel restoration; sensitivity; spectral estimate; windowing; Acoustic noise; Additive noise; Autocorrelation; Color; Colored noise; Frequency estimation; Frequency response; Image restoration; Reverberation; Signal processing;
Journal_Title :
Signal Processing, IEEE Transactions on
