to_complex_denoiser

class deepinv.models.complex.to_complex_denoiser(denoiser, mode='real_imag')[source]

Bases:

Converts a denoiser with real inputs into the one with complex inputs.

Converts a denoiser with real inputs into one that accepts complex-valued inputs by applying the denoiser separately on the real and imaginary parts, or in the absolute value and phase parts.

Parameters:

denoiser (torch.nn.Module) – a denoiser which takes in real-valued inputs.
mode (str) – the mode by which the complex inputs are processed. Can be either ‘real_imag’ or ‘abs_angle’.

Returns:

(torch.nn.Module) the denoiser which takes in complex-valued inputs.