Wavelets, Gaussian mixtures and Wiener filtering

In this work we discuss an improvement of the image-denoising wavelet-based method presented by [1]. This method is based on the estimation of the signal power at each wavelet scale and the proportion between signal and background at each scale. These parameters were estimated from the 2nd and 4th moments of the wavelet coefficients at the corresponding scale. In this work, we explored the use of any combination of the 2nd , 4th and 6th moments. We show that the 2nd and 6th moments yield better results for the experiments carried out.

• Type: Journal
• Year: 2002
• Where: SIGPRO
• Authors: Albert Bijaoui
• Keywords: Image-denoising Wavelet-based Method, Signal Power, SIGPRO 2002, Wavelet