Information fusion has been recognized as an important component in the design and implementation of multibiometric systems. Recently, improved matching and recognition performance have been reported in the literature for multimbiometric systems. Information fusion plays a major role in such systems. In fact, for multibiometric systems, information fusion can be viewed as a scheme to improve the quality of the biometric trait sample for more accurate matching performance and reduced false accept/reject rates. It is worth noting that the application of information fusion encompasses several stages in multibiometric systems ranging from raw samples acquisition to matching decision of accept or reject. In this book, we focus on the inclusion of the information fusion solution at the sample and pre-feature levels. In particular, we propose two efficient fusion schemes. One scheme is to fuse fingerprint images, and the second is to fuse iris images. The scheme of fusing fingerprint images is based on digital image watermarking, while the scheme of fusing iris images is based on a statistical modeling of the normalized iris images applied at the pre-feature level.