It's been a looong time since I was involved in a thermal printer development project... It was a Citizen LT-286, I still have a sample around...
Generally, Steve is right. You should not fire all the dots at a time, as this may lead to excess current consumption. And, guessing from the diagram, the head consists of 6 sections (64 dots each), and the odd sections are offset one microstep from the even. So you shift out your data for all 6x64 dots, then fire strobe #1, then #3, then #5; then make a microstep, and fire #2, then #4, last #6.
You should experiment with the width of the strobe, as this determines the "blackness" of the resulting dots, but this also depends on your thermal paper - you may need to implement some method how to adjust it. Don't forget to have powerful enough power supply for the case you need to print someting as a horizontal line, firing all the 64 dots at a time. If you don't run out of some fat NiCd batteries (as is a typical case for these beasts), but from AC mains, you may be well off with a relatively small transformer, if you add a big capacitor on the "head power" (separate from the "logic power", either a separate winding on the transformer, or at least a ); and implementing a "smart" algorithm, allowing to charge the capacitor between the strobe pulses.
The drive-out is probably an internal signal (there are some power transistors inside, driving the segments, driven by the strobe signals), you probably can forget about it.
For first experiments, you can probably also forget about the thermistor - unless you are going to print some very dark graphics very quickly, it should be of no issue.

Jan Waclawek