The idea is to consider each part of the ``swing'' of the input sinusoid and look at which parts of the full-wave bridge are more positive or negative, in order to figure out which way the diode conducts. We can treat the diodes in our simplest model, as one-way current valves that conduct when the back-of-the-arrow side is more positive than the vertical line side.
During the positive swing (positive hump of the sinusoid), the top
node of the bridge is more positive than the bottom. The top left
diode won't conduct but the top right one will, as it will be
forward-biased. The right node will be only 0.6 V less positive than
the top. The bottom right diode will be reverse-biased and won't
conduct. Current can flow right to left across the center resistor,
so there will be a 
to 
voltage drop across it; the current will
flow through the forward-biased bottom left diode to the bottom.
During the negative swing, it's a similar situation, except that the
current flows from the positive bottom node, through the forward-biased
bottom right diode, right to left across the center resistor, and
through the top left diode to the more-negative top node. In both cases there a to , right to left voltage drop across the center resistor. Since that's where the
output voltage is taken from, the output swing will always be positive.