Improved wavelet image denoising technique by cycle spinning and threshold selection
Aziz Khelalef, Yassine Himeur, Tahar Brahimi · The Journal of Engineering Research [TJER] · 2014
In this paper, we propose a new wavelet-based image denoising algorithm that is based on a state-of-the-art algorithm, namely FAS (Feature Adaptive Shrinkage). Two modifications are introduced in order to increase its efficiency. This consists of developing a new shrinkage function which is adapted to each level of decomposition. Also, we combined the new scheme with the cycle spinning algorithm in order to resolve the pseudo Gibbs phenomena problem. A number of experiments, carried out on various test images, demonstrate significant improvement over the conventional FAS and against other wavelet denoising methods.