A Simple and Efficient Wavelet-Based Denoising Algorithm
Using Inter- and Intrascale Statistics Adaptively
Jun Ge and Gagan Mirchandani
(Paper (pdf) / Paper (ps))
We propose a simple and efficient image denoising algorithm in the
wavelet domain. The algorithm adaptively weighs the joint inter- and
intrascale statistics of detail coefficients. Direct correlation of
detail coefficients across scales is used to select the significant
coefficients. Intrascale statistics are used to adaptively modify the
coefficients, using a new homogeneity measure. Unlike existing
algorithm using parametric models, prior knowledge and estimation of
parameters are not needed. New justification is provided for the
choice of the `most regular' wavelet derived from B-splines. The
implementation is simple and efficient, with a performance comparable
to results by state-of-art methods.