Hashing would work if the scanners were taking absolute, binary measurements without error. But they are not, not a single biometrics unit has or can have that sort of precision. If you capture your fingerprint parameters with the same device, with the same process, two or three times in a row, you'll see significant changes in the parameters from one time to the next. While the detection algorithms are designed to cope with such scanning errors, hashing would make relative comparisons fail 100% of the time.
And there lies the problem with biometrics: once you use them once (or even before you do), your "parameters" are no longer a secret under your control. If you give your fingerpring parameters to your bank, your school and your employer, each of them can in theory authenticate as you to the others.
That's why I always say: biometrics are technically useless as an authentication mechanism. They can be used for identification (replacing your username) but not for validation (your password) because they are NOT a secret, they CAN'T be revoked, you don't have the option to use different ones for different organizations and they are easy to fake. Of these issues, only the last one can be improved with better technology, the rest are intrinsic to the concept.