I am trying to figure out if there is any standard way of normalizing a sequence such as a time series against a reference sequence. I am not an expert on signal processing but I was hoping someone would be able to point out a good way of doing this. Specifically, I am working on image analysis where I measure pixel intensities along linescans. This gives me two intensity profiles along space (from 2 channels). I am trying to normalize 1 channel against the 2nd one (reference).
I assume your reference image data
y_reference is acquired when there is no light on the image sensor - that's the standard thing to do. Then the usual normalized signal is
y_normalized = (y_signal - y_reference) / y_max
y_signal is the signal image data and
y_max is the maximum possible pixel output.
y_max can be measured by increasing the light level on the image sensor until the output no longer increases.