Interestingly while we are not really capable of searching for such stars right now we have found few possible candidates [3]. I personally think these are likely to all be false positives because any civilisation capable of building a Dyson sphere will be able to capture all the energy being emitted and so the black body temperature of the sphere will be only be a little above 2.73K. These we would not be able to detect.
1. http://en.wikipedia.org/wiki/Dyson_sphere
2. http://en.wikipedia.org/wiki/Self-replicating_machine
3. http://home.fnal.gov/~carrigan/infrared_astronomy/Other_sear...
I wonder if there's reason to think that all or many stars would need capturing.
Also, is there perhaps an alternative situation, where, put simply, there power needs can be sufficiently met without capturing all that available? (Lower power tech, or some very efficient power capture / production?)
If the origin star system was far enough a way, then the slightly slower speed that the von Neumann machines would travel at should allow us to see them progressing though a distance galaxy. Such a galaxy would look to us like part of the galaxy had a chunk taken out of it. It would be worth looking through the galaxy classification data set [1, 2] to see if there are any galaxies like that out there that look like this. My feeling is we won't find any as I think intellegent life is near unique within the visible universe, but it is at least a testable hypothesis.
What I think is better proof that intelligent life is rare is the existence of stars. Stars are like finding gold coins on the street - they both only will be there if intelligent life is not.