The field of biometrics and security is a study of unique physical or behavioral traits of humans, which can be used for various applications: such as person identification, authentication, forensics, border security, attendance systems and so on. There still exist a number of problems/challenges in this field, which need to be addressed. We have addressed some of these problems: mainly deformation invariant classification of a large scale fingerprint database, touchless fingerprint recognition, scattering network for ear recognition, phase encoding of iris recognition and enhancement of latent fingerprints on banknotes. In particular, we have developed a novel algorithm for fingerprint classification, based on scattering wavelet networks, to extract translation and small deformation invariant local features. This is an example of an approach, which unifies multiresolution principles with deep learning. Automatic ear recognition system uses ear images, that are obtained from video footages or profile headshots. The process of acquisition of ear images is also nonintrusive, contactless and does not require much co-operation from the subject. The shape, structure and appearance of ears is unique and invariant to ageing. Our approach, which employs multiresolution deep learning is also useful here. A latent fingerprint on banknotes can be very useful information for law enforcement agencies, in identifying criminals in bank robbery and counterfeiters. Due to the complex background of a banknote, it is extremely challenging not only to automatic two-fingerprint matchers, but also for fingerprint experts. We have some preliminary work on analyzing latent fingerprints, which we shall briefly outline.