Rapid technology advancement and widespread use of digital imaging devices have brought about a number of forensic and provenance questions, including how an image was generated; from where an image was from; what has been done on the image since its creation, by whom, when and how. In our work, we introduce Component Forensics as a new methodology for forensic analysis of digital images and imaging devices. The proposed framework aims at identifying the algorithms and parameters employed inside various processing modules of a digital device, using only the sample data collected from device outputs without breaking the device apart. In this talk, I will illustrate the component forensic framework with our recent studies on digital cameras and present techniques to estimate the algorithms and parameters employed by important camera components such as color filter array and color interpolation modules. The estimated interpolation coefficients serve as useful features to build a robust camera identifier, which determines the camera brand/make from which an image was captured. The results obtained from such forensic analysis can also be used to quantitatively examine the similarity between the technologies employed by different camera models to identify potential infringement/licensing and to facilitate technology evolutionary studies. Building upon component forensics, we further develop a general authentication and provenance framework to reconstruct the processing history of a digital image and detect tampering operations. We model post-camera processing operations as a manipulation filter and estimate its coefficients using a linear time invariant approximation. Any inconsistencies in the estimated fingerprints from a test image, absence of in-camera fingerprints, or the presence of new post-camera fingerprints suggests that the image is not a direct camera output and has possibly undergone some kind of processing, such as tampering or steganographic embedding, after capture. To understand the fundamental performance limits of component forensics, we develop a novel theoretical framework based on estimation and pattern classification theories and define formal notions of identifiablity and classifiability of components in the information processing chain. We show that the proposed framework provides a solid foundation for information forensics to answer a number of forensic questions related to who has done what to the content, when and how.

Ashwin Swaminathan received the B.Tech degree in Electrical Engineering from the Indian Institute of Technology, Madras, India in 2003. He is currently a Ph.D. candidate at the Department of Electrical and Computer Engineering at the University of Maryland, College Park. He was a research intern with Hewlett-Packard labs (Palo Alto, CA) in 2006 and with Microsoft Research labs (Redmond, WA) in 2007. His research interests include multimedia forensics, information security, and cyber-enabled discovery.  In 2005, his paper on multimedia security was selected as the winner of the Student Paper Contest at the IEEE International Conference on Acoustic, Speech and Signal Processing (ICASSP'05). He was selected as the inaugural class of Future Faculty Fellow in 2007 by the A. J. Clark School of engineering at UMD, and received the ECE Distinguished Dissertation Fellowship award in 2008.