There are several considerations to take into account when deciding how to interpret search scores. The quality of both the gallery and probe images, the size of the gallery, the amount of human review available and the consequences of either missing a true match or accepting a false match all play a role in how the business process that uses match scores should be designed.
A common method to measure matching accuracy is the False Accept Rate (FAR). The FAR is defined as the probability that a biometrics system will incorrectly identify an individual or will fail to reject an impostor. FAR is also known as a Type II error rate. The FAR and the score provided relate to each other in accordance with a logarithmic function, log(FAR) == score, which yields interpretations such as the following: