Wavelet Denoising

  • Wavelet denoising reduces high frequency noise.
  • A direct wavelet transform is computed from the original image.
  • Noise level at each wavelet scale is estimated separately.
  • This defines a threshold for zeroing wavelet coefficients.
  • Other wavelet coefficients are shrinked according to local variance estimation (Soft thresholding, Donoho, 1995).
  • After inverse wavelet transform, the image is renormalized.