Denoising Of Digital Images Using Cyclespinning Algorithm With Shifted DWT

Noise determination and estimating a signal along with all its details proves a challenging task in signal processing. This issue has been addressed in the past using various discrete wavelet transform (DWT) based techniques. The signal is estimated as linear average of individual estimates derived from translated and wavelet-thresholded versions of a noisy signal by cycle spinning technique. In this paper, we propose a modified cycle zpinning algorithm with a new scaled down threshold of wavelet shrinkage for denoising images containing zero mean Gaussian noise using linear average of reconstructions obtained from shifted sequences’ DWT. This considerably improves the denoising performance of the conventional recursive cycle spinning algorithm and requires drasticallyless computations. Denoising performance of the proposed algorithm is benchmarked with published Recursive Cycle spinning, Buades NL means and Dual tree Complex Wavelet algorithms visually and quantitatively.