Abstract
Abstract. Spectral imaging (SI) refers to the acquisition of the three-dimensional (3D) spectral cube of spatial and spectral data of a source object at a limited number of wavelengths in a given wavelength range. Snapshot spectral imaging (SSI) refers to the instantaneous acquisition (in a single shot) of the spectral cube, a process suitable for fast changing objects. Known SSI devices exhibit large total track length (TTL), weight and production costs and relatively low optical throughput. We present a simple SSI camera based on a regular digital camera with (i) an added diffusing and dispersing phase-only static optical element at the entrance pupil (diffuser) and (ii) tailored compressed sensing (CS) methods for digital processing of the diffused and dispersed (DD) image recorded on the image sensor. The diffuser is designed to mix the spectral cube data spectrally and spatially and thus to enable convergence in its reconstruction by CS-based algorithms. In addition to performing SSI, this SSI camera is capable to perform color imaging using a monochromatic or gray-scale image sensor without color filter arrays.
Highlights
“Spectral imaging” (SI) refers to the acquisition of the threedimensional (3D) spectral cube of spatial and spectral data of a source object at a limited number of wavelengths in a given wavelength range
In order to convert a regular digital camera to an Snapshot spectral imaging” (SSI) camera for arbitrary objects, we resort here to (i) a diffusing and dispersing "phase-only" static optical element at the entrance pupil, and (ii) tailored compressed sensing (CS) methods for digital processing of the diffused and dispersed (DD) image recorded on the image sensor
We demonstrate the feasibility of reconstructing experimental SSI images with a relatively straightforward linear iterative process of "split Bregman iterations" (SBI) (Goldstein et al, 2009, Cai et al, 2009)
Summary
“Spectral imaging” (SI) refers to the acquisition of the threedimensional (3D) spectral cube of spatial and spectral data of a source object at a limited number of wavelengths in a given wavelength range. The coded aperture can be a binary mask, a gray-scaled coded mask (Rueda-Chacon et al, 2013), or a spatial-light modulator (Yuan et al, 2015) These designs yield 2D coded measurements on the sensor array, from which the spectral cube is reconstructed using CS algorithms. The limited volume of data in the DD image acquired by a 2D image sensor in a single snapshot poses a problem for the reconstruction of a 3D spectral cube. To overcome this limitation and to enable SSI, we resort to compression of spatial data in multispectral images with the aid of CS-based reconstruction algorithms. We demonstrate the feasibility of reconstructing experimental SSI images with a relatively straightforward linear iterative process of "split Bregman iterations" (SBI) (Goldstein et al, 2009, Cai et al, 2009)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
