A novel wavelet based denoising algorithm using level dependent thresholding

Wavelet domain denoising is based on the sparsity property of the Discrete Wavelet Transform (DWT) due to which information is contained in a few coefficients of high magnitude while noise is represented by a large number of small coefficients. The critically sampled DWT suffers from the drawbacks of being shift-variant and lacking the capacity to process directional information in images. Therefore the undecimated double density dual tree DWT (UDDDT-DWT) is used as it is highly directional and shift invariant. As the level of decomposition increases the number coefficients pertaining to signal increases and it is required that the number of coefficients retained be increased accordingly. The Birge Massart strategy is used therefore to compute the level dependent threshold which is applied using soft thresholding. Experimental results over a range of noise variances indicate that the proposed method performs better than other state of the art methods considered.