The short question is: Can it work?
The short answer is no.

The reason is that your uC can not handle the speed with which pulses come in when the encoder stop exctly across a boundary in the encoder. I have seen pulses coming at several MHz from an encoder connected to a 5 RPM unit.

Agilent nee HP has a very good chip that resolves this, I have, in the past, resolved it in a CPLD. If you draw a diagram of all the reversals (I hope some day one of those currently working with encoders will take the time to do so and publish it) you will see that the a=clock, b=direction fails in 2 of 16 cases.

Erik