FPN isn't that big a deal, because it's fixed. Dark image subtraction will take care of it. Depending on what you're designing your image sensor for (scientific measurement, low-light applications, portrait photography etc) different things will be a priority - noise, sensitivity, dynamic range etc.
As mentioned a source follower will do the job, also if you have the area available use a differential pair (better linearity). You'll also need to design an op-amp to drive any voltages off chip. Obviuosly the first thing this voltage wants to see when it gets off-chip is the input to another buffer.
APS sensors are fine, but if you're looking at high frequency stuff, then they are useless and you'll need to use a log-pixel (which also have a huge dynamic range, if that is important)