poisson

dnois.sensor.poisson(signal: Tensor, a: float | Tensor, noise_only: bool = False) → Tensor

Applying poissonian noise to signal:

\[\tilde{\mathbf{x}}\sim\mathcal{P}(\mathbf{x}/a).\]

Attention

This is a non-differentiable operation.

Parameters:

  • signal (Tensor) – Input signal.

  • a (float | Tensor) – Scale factor \(a\) in Poisson distribution. Must be broadcastable with signal if a Tensor.

  • noise_only (bool) – If True, return noise rather than noisy signal. Default: False.

Returns:

Noisy signal.

Return type:

Tensor