pattern noise subtraction removed mean background signal

[eugene]

the pattern noise subtraction code results in a zero mean background. for cells. as a result, the background statistics from the chip stage reported to the ipp database are meaningless. we probably want to either add back in a constant mean value for the cell, or maybe even a low-frequency y-direction model.

[Paul Price]

I think the psphot background subtraction does the same thing, and is required.

Should we simply do the background statistics before we do background subtraction?

[eugene]

this was fixed when the pattern row & pattern cell corrections were recently (Jan 2010) re-worked. the median cell value is added back to the cell after the rows are subtracted.