ImageFlatteningNoiseApprox¶
- class giant.image_processing.ImageFlatteningNoiseApprox(value, names=None, *, module=None, qualname=None, type=None, start=1, boundary=None)[source]¶
This enumeration provides the valid options for flattening an image and determining the noise levels when identifying points of interest in
ImageProcessing.find_poi_in_roi()
You should be sure to use one of these values when setting to the
image_flattening_noise_approximation
attribute of theImageProcessing
class.- GLOBAL = 'GLOBAL'¶
Globally flatten the image and estimate the noise level from it.
In this the image in flattened by subtracting a median filtered version of the image from it and a single noise level is approximated for the entire image either through sampling or through the
dark_pixels
of the image.For most OpNav cases this is sufficient and fast.
- LOCAL = 'LOCAL'¶
Locally flatten the image and estimate the noise levels for each local region
In this the image in flattened by splitting it into regions, estimating a linear background gradient in each region, and the subtracting the estimated background gradient from the region to get the flattened region. An individual noise level is estimated for each of these regions through sampling.
This technique allows much dimmer points of interest to be extracted without overwhelming with noise, but it is generally much slower and is unnecessary for all but detailed analyses.