Currently a single frame is collected to serve as light and dark reference imagery.
Analysis of the dark ref imagery indicates frame-to-frame noise with a standard deviation ~5% f.s. which means the normalized imagery will contain (non-iid) noise to ~15% f.s. which is rather a lot.
To reduce the noise in the final image we should collect multiple frames and average them before using them to normalize the RGB imagery. Since the noise will reduce proportional to sqrt(N), collecting 25 images should be sufficient to reduce the noise to <1% f.s
Ideally the number of .images to collect would be runtime configurable.
Any other sensor that currently collects a single frame (e.g. MLT) should be modified to allow multiple to be collected.
DB0077 [Created: 2016-12-08]
Currently a single frame is collected to serve as light and dark reference imagery.