Hi Herbert,
Scale factors and zero offsets are applied to make input frames statistically compatible. Scale factors compensate for differences in statistical dispersion (scale differences), and zero offsets compensate for different central values (location differences). Simplifying a little,
scale can be understood here as the width of the main histogram peak, and
location as the position of the main histogram peak. These differences must be removed in order to be able to compare all input frames statistically, which is crucial for the task of pixel rejection.
Weights are multiplying factors to account for signal-to-noise ratio differences in an optimal weighted mean combination.
Scale and location estimates are, respectively, robust estimates of statistical scale (dispersion) and location (robust mean) for the output integrated image. SNR estimates are signal-to-noise ratio estimates. SNR is a non-robust statistical function, so these values must be taken with at least one (or two, or three...) grain of salt. Basically, SNR estimates are given exclusively for informational purposes,
never to make consistent decisions. Always use effective noise reduction estimates to compare different integration results.
See the
ImageIntegration reference documentation for detailed and more in-depth descriptions.
