it's extremely unlikely that you will be able to make this work...
a) Capacitive sensing is only reliable with very good engineering
b) Capacitive sensing is generally very finicky and unreliable
c) Capacitive sensing without adequate ( i.e. 100K or less) pull-ups or pull-down resistors is inherently unreliable as the pins are not tied to a certain state; i.e. you will rely too much on the local electromagnetic conditions to set the "off" state.
d) You're basing it on the Arduino capacitive sensing idea; the CMOS 4051 is another type of chip architecture, with a different type of logic inputs
e) you're multiplexing the inputs, which will probably cause havoc with the sensitive levels required to sense capacitively
f) doing it correctly will require very good shielding and PCB design to get it to be reliable.
I could write a lot more reasons here, but trust me, it isn't going to work the way you want, or at all, so
don't waste your time
Sure you could spend a few months and maybe get it working semi-reliably, but why bother when there is a 4$ solution waiting around the corner that is known reliable?You can get everything you want already engineered in a multi-channel QTprox touch-sensor chip if you want to do capacitive sensing...