During the training step for each person a biometric template, also called prototype, is calculated. It is generated from the preprocessed data (features). During the classification of a sequence its features are being matched with this biometric template.

The template is made up of a separate part for each classification trait and can be understood as a compact representation of the collected feature data, where useless or redundant information is discarded.

Independent of the number of training sequences, each part consists of a fixed amount of data representing each person’s characteristics. In addition to this, traits relying on temporal information, e.g. voice and mimic, may need to store some extra data for each training sequence to keep information about the time relations.

Size of the biometric template

Considering 5 training patterns, the resulting template contains 3 950 Bytes of biometric data. The template is packaged into a data structure that contains class and creation time information as well as structure data and version numbers. The additional information extends the size up to 5 212 Byte (pure data block, signed).