is related to the pulses coming when the encoder 'vibrates' while positioned across a boundary between opaque and clear on the optical disk.
Hmm, I see then.
BTW, what is about another software way: just ignore all pulses which income very frequently. First of all, we need to calculate maximal frequency of pulses when wheel rotates with maximal possible speed. Let say, it is 1ms. If we detect a pulses` sequences which income more fast (each-by-each time is shorter than 1ms) so we just drop this fact and do nothing.

Regards,
Oleg