How to create a flawless biometric test?

Question

Biometric identification

We plan to create a project in which each user account corresponds to a unique real person. One entry per living person, not a phone or email. We need a biometric test that meets the requirements:
1. The test can only be valid for representatives of the Homo Sapiens species (prints and voice of a cat, dog, etc. are rejected)
2. The test cannot be passed using a mockup or imitation
3. Valid result The test cannot be reproduced with a sample in hand.
4. The test cannot be passed with fictitious parameters.
5. According to the test result, it is impossible to establish other personal data about a person (what is his name, where he lives, etc.), except that he is the same person as %USERID% or new.
Most likely, a regular handwritten signature will do: https://habr.com/post/189848/
We are not afraid of difficulties. I leave you the right to consider the customer crazy, who has extra money for development. It just needs to work. Interested in any of your ideas. Thanks in advance!

Answer 1

We are not afraid of difficulties.

but why be afraid, on the toaster they will give a ready-made solution right away))
why not have a competition with an incentive reward?
The question is not so simple as to sit down and throw it on your knee and / or even more so tell some unique solution for free.

Answer 2

Combination.
All individually - forged.
And the signature, and prints, and even the timbre of the voice.
Combination is more difficult.

Answer 3

Read with the customer to start "Personal Data Law"

Answer 4

Scull. Its form changes little over time and is unique to each person, although collisions are possible.
Accurately scanning the shape of the skull over hundreds of points could be up to the task.
I wanted to go to the bank - do an X-ray of the head until you start glowing in the dark) In general, the bones are clearly visible on the MRI. Or you can try to analyze the sound conduction of the skull.

Answer 5

Simple and tasteful:
1. WebRTC: we display a piece of the work / text and ask you to read it in front of the webcam (20-30 seconds) into the microphone before each work session (logon).
2. And then - recognize the video and determine the userID.