In this concept, a biometric sample which is represented in a binary string b, is added to a random codeword cw of an error correction method to calculate the commitment function c by:
c = cw + b
At the same time, the values of c and the hash value of cw are stored somewhere for authentication purposes. On decommitment, the codeword cw´ is regenerated by:
cw´ = b´ - c
where b´ is the binary string of the test biometric sample, and the verification is successful if the hash value of cw´ equals the stored hash value of cw. Regarding the codewords with a minimum Hamming distance d, the authentication would be successful when the Hamming distance between b and b´ is lower than or equal to d/2. Note that the commitment term refers to the fact that the error correction codes are committed to remove the biometric variabilities on authentication.
I would say this approach can be used for user authentication in the first place. It could not probably be used for encryption aims, where it cannot generate the same binary string at all times due to the biometric variances.
Hope this helps.
- Badges
- Science topic
- Similar topics
- Statistics
- Survey