For series resistance, select a points just above the knee point on two of the I(V) curves at different irradiance levels. Let the two points be (V1, I1) and (V2, I2). Then (V2-V1) / (I1-I2) is an estimate of the resistance. This can be repeated at different irradiance levels as many times as the estimate is required.
For shunt resistance, simply estimate the slope of any I(V) curve at the short circuit point (V = 0). This slope is equal to -1/Rsh.
These are standard experimental procedures, and should be available in many books.
Sanjay is quite correct - of course! Basically the slope of the IV curve at V ~ 0 is a measure of the shunt resistance and at I ~ 0 a measure of the series resistance. Incidentally, the point where I/V = - dI/dV is the maximum power point. Maybe some MPPT's work on this principle, though I haven't come across one.
What the "PV community" refers to as MPPT by incremental admittance is exactly what you have suggested. Of course the practical realisations have the usual experimental approximations involved.
If you don't want to or cannot conduct experiments to determine the parameters, you can still calculate the model parameters using only the PV panel datasheet values. For details, please have a look at: