Denoising NMR data

Due to rapidly changing magnetic fileds NMR signals are often noisy.
Haar wavelet transform is used to decompose the signal into subbands. Each subband is thresholded adaptively to suppress noise

The recursive decomposition employs the signal to noise ratio (SNR) as the objective function instead of entropy. The different thresholds are applied to the subbands and valid frames are extracted and reconstructed to get the denoised NMR data.