SIAM Conference on Imaging Science (IS22)

Slides from Mini-symposium on Mathematical Methods for Computational Sensing and Imaging

Overview of the Session

Traditional sensing or imaging methods are pivoted on a “capture first, process later” philosophy. Here, the recovery/restoration is typically decoupled from the acquisition. This fundamentally limits the applicability of mathematical algorithms because if the sensors result in information loss of any form, the algorithms may no longer be able to achieve their guaranteed performance.

In contrast, the emerging field of Computational Sensing and Imaging (COSI) is based on a joint design of hardware and algorithms; one relies on a carefully optimized capture process yielding encoded measurements from which the information is decoded using mathematically guaranteed recovery algorithms. In recent years, the COSI approach has enabled a number of breakthroughs and continues to inspire new research.

The goal of this session is to bring together interdisciplinary efforts that push frontiers by exploiting a co-design of novel acquisition and mathematical methods. Thus inspiring further investigation into mathematical algorithms. The session is composed of 12 speakers focusing on 3 different aspects of COSI. The talks focus on thematic subtopics that (a) inspire new acquisition, (b) showcase unconventional imaging frontiers and (c) demonstrate new algorithmic capabilities.

By bringing interdisciplinary efforts together, we hope to inspire an end-to-end thinking where mathematical ideas (a) facilitate new algorithms and (b) help conceptualize novel hardware/acquisition.