About the book
Optical images could be acquired either directly by using, e.g., a simple lens, or indirectly, as in the case of tomography, by using data acquisition hardware followed by a computational image reconstruction step. Computational optical imaging is a highly promising approach for direct or indirect imaging that involves the joint design of image acquisition hardware and digital processing algorithms to achieve imaging performance that would otherwise be unattainable by conventional systems.
Despite much advances in computational power, the processing (and storage) of acquired signals in applications such as real-time medical imaging, remote surveillance, and spectroscopy still pose a remarkable challenge. In addition, it may be too costly, or even physically impossible, to build optical hardware to acquire imaging data at the Nyquist sampling (measurement) rate that is required for high imaging performance, i.e., high resolution, large field-of-view, and high frame rates.
Compressed Sensing involves the digital construction of an image using a number of samples (measurements) that are significantly less than its dimension. By assuming that the unknown image is sparse in the domain where the measurements were acquired, one could use this sparsity constraint as prior information to obtain an approximate but accurate reconstruction of the image from relatively few samples.
This book will be about the design and implementation of optical imaging systems, particularly, optical microscopy, optical coherence tomography, and hyperspectral imaging, using compressed sensing approaches. It will be particularly concerned with compressed sensing data acquisition approaches, and computational challenges involved in processing very large imaging data sets that traditionally result from 3D and/or real-time biomedical and industrial imaging applications.