Revision history [back]

The traditional method of adding Poisson noise assumes that the image data is in integer form. However, if the image data is not in integer form, the following approach can be used:

1. Normalize the image data to a range of [0, 1].
2. Scale the image data by a factor of 10^x, where x is the number of decimal places in the image data.
3. Apply Poisson noise to the scaled image data using a Poisson distribution with λ equal to the mean of the scaled image data.
4. Scale the Poisson noise back by a factor of 10^x.
5. Denormalize the data back to its original range.

Note that this method is an approximation and may not produce accurate results for very small or very large values of image data.