On a different level but it might be an anology with pmrl signal processing:http://www.iop.org/EJ/abstract/1464-4258/5/5/358
In this link an optical system is described where a special designed lens systems alters the captured image into an intermediate image instead of the negative or positive image we are used to see with photograpy.
This intermediate image looks blurry (no pun intended) and the captured image is not recognizable at all.
However applying dsp filter functions to the intermediate image results in a photograpic image that has a greatly increased depth of field over a normal captured image.
Now for the anology:
Think of the signal coming of a triple layer disc as the intermediate image and the dsp as the used pmrl technology. The end result is of coarse the output signal from the layer that is read.
The signal from the triple layer disc is contains the reflections of two out of focus layers resulting in a blurry and jittery signal (call not random noise) plus the signal of the layer that is focussed upon.
My understanding of this is that the focus layer signal is modulating the signal from the out of focus layers.
The pmrl technology can be used to filter out the correct signal.
A revised function is required because focussing on each layer requires different pattern recognition functions because of the differences in reflectivity for each layer.
Worst case is the 'top' layer with weak reflection from the layer and 'low' noise from other out of focus layers.
Best case is the 'bottom' layer with high reflection and 'high' noise from the out of focus layers with the 'top' layer being to most out of focus. (I suspect that reading the middle layer might actually be the most difficult layer to readout as the surrounding layers are equally out of focus. )
With current dsp technology extracting signals burried in noise is done in many applications already. SA is just a factor affecting how the noise patterns are shaped.