Coded aperture design in compressive spectral imaging

Arce, Gonzalo R.Arguello-Fuentes, HenryCompressive Spectral Imaging (CSI) systems sense 3D spatio-spectral data cubes through just few two dimensional (2D) projections by using a coded aperture, a dispersive element, and an FPA. The coded apertures in these systems, whose main function is the modulation of the data cube, are often implemented through photomasks attached to piezoelectric devices. The optimization of such coded aperture patterns is an actual area of research. Two remarkable improvements on this configuration have been recently proposed. First, the replacement of the photomask by digital micromirror devices (DMD) for block-unblock coding in order to facilitate the capture of multiple projections/snapshots or the capture of multiple shots at video rates without the displacement of the optical elements on the system. Secondly, the replacement of block-unblock coded apertures by patterned optical filter arrays, referred as “colored” coded apertures, which not only allow spatial modulation but spectral modulation as well. Despite the improvements, the design of the coded aperture patterns is still constrained by hardware considerations. This dissertation aims to overcome these hardware considerations by developing different coded aperture design strategies. ☐ When using the DMD for coding the data cube, the DMD resolution and the possibility to use multiple shots have to be considered. Usually, the pitch size of the DMD mirrors is different than the pitch size of the pixels in the detector. The mismatch of the DMD mirrors and the detector pixels is such that pixel-to-pixel correspondence is not achieved. The first proposed strategy is a mismatching coded aperture design to exploit the maximum resolution of the coding element and the detector. Additionally, the capture of multiple snapshots could be highly exploited to extract prior-information of the scenes, here a second strategy is proposed, the use of side information in CSI not only to improve the reconstructions but to design scenes-adaptive coded aperture patterns. ☐ On the other hand, when using “colored” coded apertures, its real implementation in terms of cost and complexity, directly depends on the number of filters to be used as well as the number of shots. A shifting color coded aperture optimization featuring these observations is proposed as the third strategy with the aim to improve the quality reconstruction and to generate an achievable optical implementation. ☐ The mathematical models of the different strategies of computational imaging to overcome the limitations of actual CSI systems will be presented along with testbed implementations. Simulations as well as experimental results will prove the accuracy and performance of the three proposed coding strategies.University of Delaware, Department of Electrical and Computer EngineeringPh.D