What would you advice me, what type of specimens do I need to take SEM pictures of petrified wood. do I need the polished samples, rough specimen or thin section?
You probably need to do several trials before establishing a correct procedure but regarding visual properties of petrified wood in you images I suggest to start with rough specimen.
Can't know for sure for petrified wood (never came across one sample of this type) but for wood, the answer to your question depends on what you want to observe. From the picture you're showing, I'd try two sample preparations : freeze in LN2 and fracture or simply cut with a sliding microtome using a brand new razor blade or metal knife (or glass knife if you've got one). Both techniques will produce a block face you can image.
Make sure you either use variable pressure in the SEM or sputter largely prior to observations if using a HVac SEM.