Colloquium: Mark A. Neifeld

    Date: 
    Thursday, January 30, 2014 - 3:30pm - 5:00pm
    Location: 
    Meinel 307
    Description: 

    "Thoughts on Computational/Compressive Imaging"

    Abstract(s): 

    Compressive imaging exploits the redundancy present in natural scenes in order to reduce the number of samples required for image reconstruction. The seminal work of Donohoe and others have provided rigorous mathematical results relating the redundancy of the scene (i.e., signal sparsity) to these new compressive sampling requirements, clearly demonstrating that sub-Nyquist sampling is almost always possible with little or no loss of image quality. On its surface, this approach appears to offer a path to substantially reduced camera cost. If samples require measurement resources (i.e., so the argument goes), then fewer samples must correspond to fewer resources and lower cost. Despite the obviousness of this argument, compressive imaging continues to be more of a curiosity than a practical method with widespread utility. In this presentation I will discuss three possible reasons for this disconnect:

    1. the photon count constraint,
    2. the use of random projections and
    3. requirements on imaging optics.

    The presentation will begin with an overview of compressive imaging and its mathematical underpinnings. This discussion will be followed by a taxonomy of optical approaches to compressive imaging along with the physical principles underlying each. The taxonomy presented here will aid in the discussion of Item 1 above. Sequential, parallel and hybrid optical architectures will be discussed and the signal-to-noise scaling of each will be quantified. Recent progress away from random projections will also be discussed. This recent work is seen to offer some promise for addressing Item 2; however, all of these solutions require an optical image to be formed and so do not offer any reduction in optical hardware complexity. Item 3 therefore represents a significant unaddressed challenge for compressive imaging.

    Speaker Bio(s): 

    Mark A. Neifeld received his Ph.D. from the California Institute of Technology in 1991 and has since been a faculty member in the University of Arizona's College of Optical Sciences and department of electrical and computer engineering. He was also recently a program manager at the U.S. Defense Advanced Research Projects Agency's Defense Sciences Office, where he started programs on quantum information, compressive sensing and computational imaging. He has coauthored more than 120 journal articles and 250 conference papers in the general areas of optical physics and engineering. Neifeld is a fellow of both the Optical Society (OSA) and SPIE and a member of the American Physical Society and IEEE. He has served on the organizing committees of numerous conferences and symposia. He has been a two-term topical editor for Applied Optics and a three-time guest editor of special issues of Applied Optics. His current research interests include computational sensing, multimode optical communications, optical orbital angular momentum, 3-D optics and multidomain optimization.