The optional string defaults to LInear results in a linear scaling. If you specify LOg or Sqrt, the linear transformation is applied first and then the specified function applied.
Normally the full image is scaled unless you the Reg option.
This command first subtracts ioff*ifac*ifac from the data and then multiplies by iscale. Basically this allows the same offset to be used even if the data are read with different compressions.
data = iscale*(data-ioff*ifac**2)
Note, ioff can be the string MIn or MAx, in which case the data minimum or maximum is used. Thus after applying 'IScale MIn' the new data minimum value will be zero.