Add quantization noise + remove sky_image quantization #100
Conversation
Quantization noise
|
Question: |
euclidlike_imsim/noise.py
Outdated
| quantization_noise.fill(0) | ||
| quantization_noise.addNoise(galsim.DeviateNoise(galsim.UniformDeviate(rng))) | ||
| noise_img += quantization_noise | ||
| noise_img -= 0.5 |
There was a problem hiding this comment.
Where does this subtraction by 0.5 come from?
There was a problem hiding this comment.
With quantization we assume that if you have a value of 100.5 ADUs it has the same chance to round to 100 or 101 which lead to +/-0.5. The Uniform distribution from GalSim return value in the range [0, 1], the -0.5 is just to rescale the range to [-0.5, 0.5].
Co-authored-by: Rachel Mandelbaum <[email protected]>
|
|
@rmandelb Ready to for review |
| if cfg_noise["quantization_noise"]: | ||
| quantization_noise = noise_img.copy() | ||
| quantization_noise.fill(0) | ||
| quantization_noise.addNoise(galsim.DeviateNoise(galsim.UniformDeviate(rng))) |
There was a problem hiding this comment.
On the github interface something funny is happening with indentation? Is this a tabs vs. spaces thing or just a fluke?
| # Note that the image type after this step is still a float. If we want to actually | ||
| # get integer values, we can do new_img = galsim.Image(full_image, dtype=int) | ||
|
|
||
| # Here we add quantization noise to the image. This prevent to have problems due |
There was a problem hiding this comment.
| # Here we add quantization noise to the image. This prevent to have problems due | |
| # Here we add quantization noise to the image. This prevents problems due |
Co-authored-by: Rachel Mandelbaum <[email protected]>
Co-authored-by: Rachel Mandelbaum <[email protected]>
Co-authored-by: Rachel Mandelbaum <[email protected]>
Co-authored-by: Rachel Mandelbaum <[email protected]>
Co-authored-by: Rachel Mandelbaum <[email protected]>
Co-authored-by: Rachel Mandelbaum <[email protected]>
Add quantization noise to the image. This allow to improve background estimation, see #85.
Below are the results with the quantization noise on. It looks like the original signal is not impacted.