A method for optimizing initial discriminating information extracted from a two-dimensional image of the patient dental arches, referred to as “acquired image”, by a three-dimensional digital reference model of at least one portion of a patient arch, the method including the steps: C1. assessing quality of the initial discriminating information and quality threshold, filtering to keep only the initial discriminating information that has quality higher than the quality threshold, and defining “the discriminating information to be tested” as the initial discriminating information selected; C2. testing consistency between the discriminating information to be tested and reference model; C3. assessing test result and, in accordance with the assessment: adding discriminating information that was not kept to the discriminating information to be tested and/or deleting discriminating information from the discriminating information to be tested, and then returning to step C2. or; defining the optimal discriminating information as the discriminating information to be tested.