sorry signal - just read the post I missed earlier about them wanting to convey better security (even if it is a false sense)...
If you/they want to go down the biometric path, I have a lot of experience there... Optical recognition works the best as far as unique identification, but it's easy to fake... Capacitive isn't as dead-on and is prone to temperature differences, etc - but seems harder to fake with copies. I've even done testing with a finger-vein reader... you're not faking that for anything, but it was a PITA to get positioned just right.
For something with high traffic and variables, if you really want Biometrics, I'd probably suggest the Ingersol Rand Hand Scanners - they take measurements of your hand - length of fingers and width in a 3D model - they adjust for things like temperature where your hands swell a little or shrink - and once you're used to them, they're easy and reliable - they're just a bit larger than little fingerprint readers. I even have 4 of them laying around I'm not using - they have custom software on them, but if you find someone willing to work with them, I'll send you one for free (seriously - and if you use it and get it up and running, I'll send a 2nd for a shelf backup). I tested them but never deployed them because of unique network connectivity issues (very unique to my company).
That said, I still also fully stand by my above comments - this is all a false sense of security - just a better one... the above issues still exist and you're still really no more secure.
If you want real security, you need one-way paths for entry/exit - so someone leaving doesn't hold the door for someone entering - you need a good man trap, and you need someone on duty who can override a valid access-request because someone following them looks suspicious - someone watching at all times. Technology alone can't solve this.