In response to #5314
Thanks for your comment @kalemwo1994 ... and for further picking my curiosity about software binning. As a result I run a couple of tests to evaluate their effects on SNR.
Software used:
- Integration: APP
- Analysis and Resampling: PI
Methods and Results:
First I integrated the frames as they are (bin1) in APP. The SNR is 39.76 db. If I then IntegerResample (Resample factor = 2) this image, the SNR becomes 44.48 db. The overall best SNR however is the result of a Resample at 50% of this, i.e. 46.80 db.
I then loaded all raw frames into PI and performed an IntegerResample (RSf=2) on each of them (tedious) before integrating them with APP. The SNR is now 45.55 db. (The average SNR of the raw frames is 24.42 ± 0.38, the average SNR of the IntegerResampled frames is 30.29 ± 0.30.)
When I integrate the frames in APP and use a 0.5 scaling during integration, I get an SNR of 45.32 db.
Conclusion:
The best overall SNR comes from Resampling (not IntegerResample) the raw bin1 integrated image in PI with a ratio of 50%.
Using a 50% scaling during APP integration yields a result very close to integrating individually IntergerResampled (RSf=2) frames. In both cases, it is better than to IntegerResample the bin1 integrated image by about 1 db and a bit more than 1 db worse than a 50% Resample of the bin1 integrated image.
So overall, using the Resample method in PI yields a Amplitude ratio improvement of 30% over IntegerResample and an improvement of 18% versus scaling during APP integration.
What remains to be evaluated, is the effect of these differences on image processing following channel integration.
Cheers
Manuel