Hessian Denoising#

API#

pycudasirecon.hessian_denoise(initial, iters=50, mu=150, sigma=1, lamda=1)[source]#

Hessian (Split-Bregman) SIM denoise algorithm.

May be applied to images images reconstructed with Wiener method to further reduce noise.

Adapted from MATLAB code published in Huang et al 2018. See supplementary information (fig 7) for details on the paramaeters.

Parameters:

  • initial (np.ndarray) – input image

  • iters (int, optional) – number of iterations, by default 50

  • mu (int, optional) – tradeoff between resolution and denoising. Higher values result in increased resolution but increased noise, by default 150

  • sigma (float, optional) – When structures change slowly along the Z/T axis, use a value between 0 and 1 to obtain trade-off between effective denoising and minimal temporal blurring. Set to zero to turn off regularization along the Z/T axis. by default 1

  • lamda (int, optional) – [description], by default 1

Returns:

result – denoised image

Return type:

np.ndarray

References