Incorporating Ancillary Information in Multibiometric Systems

Information fusion in a multibiometric system can be accomplished at the sensor, feature, match score, rank, or decision levels. Depending on the level of fusion, inputs to the fusion module may consist of raw images, features, match scores, ranks or identity decisions generated by the individual biometric sources. Apart from these inputs, a multibiometric system may have access to ancillary information that may be beneficial in the decision generation process. Intrinsic ancillary information is derived from the same biometric sample that is used for verifying or establishing the identity of the user. An example of intrinsic information is the quality of the acquired biometric sample (e.g., fingerprint image quality). Extrinsic information is derived from sources other than the acquired biometric sample. For instance, characteristics such as gender, ethnicity, height or weight of the user (collectively known as soft biometric traits) can be obtained as the user approaches a fingerprint recognition system. Though the ancillary information may not be directly related to the identity of the user, it is still useful for recognition in many ways, especially in a multibiometric system. The main difficulty in incorporating ancillary information in a multibiometric system lies in (i) designing techniques that can automatically extract the required ancillary information from the individual, and (ii) designing fusion mechanisms that can effectively utilize this additional information to improve recognition accuracy. This chapter presents some techniques that have been proposed in the literature to address these challenges in the context of quality-based fusion and soft biometrics.