|
A few thoughts
After stacking PI produces a 32 bit FP image. It can produce a 64 bit one.
All further processing was done in 32 bit linear form ie unstretched.
Why is this important?
All the dim nebulousity is in about the bottom 2% of the 16 bit image. This at best is about 1300 levels (0.02x65000) which is 10 or 11 bits.
By stacking many images into a 32 bit floating point image this bottom 2% now has many more levels available. Millions in fact.
When we now apply DBE Dynamic Background Extraction and ATW Atrous Wavelet Transform to this 32b FP data the result is far more mathematically valid.
Without DBE and ATW this very dim data would not even appear, let alone almost noise free.
ATW is the best method I have seen to reduce noise in the dark background without affecting ANY signal.
More on this later.
Bert
Last edited by avandonk; 06-06-2014 at 08:59 AM.
|