P.L. Dragotti and M. Vetterli. Shift-invariant gibbs free denoising algorithm based on wavelet transform footprints. In SPIE International Symposium on Optical Science and Technology, San Diego, California (USA), July 2000.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....they are still highly visible. The method we are presenting in this paper performs a far better artifact reduction, so that in most cases the pseudo Gibbs phenomena simply vanish. Let us mention another approach of artifact free wavelet denoising recently introduced by Dragotti and Vetterli in [6]. These authors propose to apply a vector thresholding in place of the classical scalar thresholding, and to replace remaining vectors by the closest footprint obtained as the significant wavelet coefficients generated by singularities of piecewise polynomial functions. In this way, they ensure a ....

P.L. Dragotti and M. Vetterli, "Shift-invariant gibbs free denoising algorithm based on wavelet tranform footprints," in Proc. of SPIE'2000, Wavelet Application in Signal and Image Processing, 2000.

No context found.

P.L. Dragotti and M. Vetterli. Shift-invariant gibbs free denoising algorithm based on wavelet transform footprints. In SPIE International Symposium on Optical Science and Technology, San Diego, California (USA), July 2000.

No context found.

P.L. Dragotti and M. Vetterli. Shift-invariant gibbs free denoising algorithm based on wavelet transform footprints. In SPIE International Symposium on Optical Science and Technology, San Diego, California (USA), July 2000.

....appear only around discontinuities. Moreover, if N is the maximum degree of any polynomial in the signal, then the wavelet coefficients generated by any discontinuity lie on a N 1 dimensional sub space represented by the elementary footprints f (i) k . Based on this consideration, in [4] it has been proposed to denoise the signal directly in the footprint domain. The advantage of doing so is that the dependency between wavelet coefficients is exploited. Let us call F the noisy version of X . The non adaptive denoising algorithm proposed in [4] works this way: 1. Compute a J ....

....k . Based on this consideration, in [4] it has been proposed to denoise the signal directly in the footprint domain. The advantage of doing so is that the dependency between wavelet coefficients is exploited. Let us call F the noisy version of X . The non adaptive denoising algorithm proposed in [4] works this way: 1. Compute a J level wavelet transform of the noisy signal: G = WF . 2. Define a threshold Th = oe p 2 ln T . 3. For each possible discontinuity position k 2 [1; T Gamma1] compute the N 1 inner products G; f (i) k , i = 0; N . 4. Choose the location k1 such ....

P.L. Dragotti and M. Vetterli "Shift-Invariant Gibbs Free Denoising Algorithm based on Wavelet Transform Footprints ", In Proc. SPIE

....(SNR=19.0) c) Denoised Signal using footprints (SNR=22.6dB) d) Denoised Signal with Soft Thresholding (SNR=21.3dB) algorithm outperforms the Soft Thresholding algorithm by about 3dB. However footprint denoising can be done without compression. Further results on this topic can be found in [2]. Finally we would like to know if these results can be extended to real world signals like images. At the moment it is not clear how to extend the notion of footprint to the two dimensional nonseparable case. However if we decide to compress each line of an image independently then the results ....

P.L. Dragotti and M. Vetterli "Shift-Invariant Gibbs Free Denoising Algorithm based on Wavelet Transform Footprints ", Accepted for SPIE

No context found.

P.L. Dragotti and M. Vetterli, "Shift-invariant gibbs free denoising algorithm based on wavelet tranform footprints," in Proc. of SPIE'2000.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC