deposit_your_work

Wavelet-based deconvolution for ill-conditioned systems

Files in this item

Files Size Format View
Nee2000Feb1Waveletbas.PS 407.2Kb application/postscript View/Open

Show full item record

Item Metadata

Title: Wavelet-based deconvolution for ill-conditioned systems
Author: Neelamani, Ramesh; Choi, Hyeokho; Baraniuk, Richard G.
Type: Journal Paper
Keywords: Temporary
Citation: R. Neelamani, H. Choi and R. G. Baraniuk, "Wavelet-based deconvolution for ill-conditioned systems," IEEE Transactions on Image Processing, 2000.
Abstract: We propose a hybrid approach to wavelet-based deconvolution that comprises Fourier-domain system inversion followed by wavelet-domain noise suppression. In contrast to other wavelet-based deconvolution approaches, the algorithm employs a regularized inverse filter, which allows it to operate even when the system is non-invertible. Using a mean-square-error (MSE) metric, we strike an optimal balance between Fourier-domain regularization (matched to the convolution operator) and wavelet-domain regularization (matched to the signal/image). Theoretical analysis reveals that the optimal balance is determined by the Fourier-domain operator structure and the economics of the wavelet-domain signal representation. The resulting algorithm is fast (O(N\log N) complexity for signals/images of N samples) and is well-suited to data with spatially-localized phenomena such as edges and ridges. In addition to enjoying asymptotically optimal rates of error decay for certain systems, the algorithm also achieves excellent performance at fixed data lengths. In real data experiments, the algorithm outperforms the conventional time-invariant Wiener filter and other wavelet-based image restoration algorithms in terms of both MSE performance and visual quality.
Date Published: 2000-02-01

This item appears in the following Collection(s)

  • ECE Publications [1047 items]
    Publications by Rice University Electrical and Computer Engineering faculty and graduate students
  • DSP Publications [508 items]
    Publications by Rice Faculty and graduate students in digital signal processing.