In this paper, we present a new hyperspectral compact camera which is designed to have high spatial and spectral resolutions, to be vibrations tolerant, and to achieve state-of-the-art high optical throughput values compared to existing nanosatellite hyperspectral imaging payloads with space heritage. These properties make it perfect for airborne and spaceborne remote sensing tasks. The camera has both hyperspectral and panchromatic imaging capabilities, achieved by employing a wedge-shaped liquid crystal cell together with computational image processing. The hyperspectral images are acquired through passive along-track spatial scanning when no voltage is applied to the cell, and the panchromatic images are quickly acquired in a single snapshot at a high signal-to-noise ratio when the cell is voltage driven.

Liquid crystal phase retarders are utilized by photonic devices and imaging systems for various applications, such as tunable filtering, light modulation, polarimetric imaging, remote sensing and quality inspection. Due to technical difficulties in the manufacturing process, these phase retarders may suffer from spatial non-uniformities, which degrade the performance of the systems. These non-uniformities can be characterized by measuring the spectral transmission at each voltage and each point on the liquid crystal cell, which is time consuming. In this work, we present a new fast and simple method for measuring and computationally estimating the spatial variations of a liquid crystal phase retarder with planar alignment. The method is based on measuring the spectral transmission of the phase retarder at several spatial locations and estimating it at others. The experimental results show that the method provides an accurate spatial description of the phase retarder and can be employed for calibrating relevant systems.

The utilization of compressive sensing (CS) techniques for hyperspectral (HS) imaging is appealing since HS data is typically huge and very redundant. The CS design offers a significant reduction of the acquisition effort, which can be manifested in faster acquisition of the HS datacubes, acquisition of larger HS images and removing the need for postacquisition digital compression. But, do all these benefits come at the expense of the ability to extract targets from the HS images? The answer to this question, of course, depends on the specific CS design and on the target detection algorithm employed. In a previous study we have shown that there is virtually no target detection performance degradation when a classical target detection algorithm is applied on data acquired with CS HS imaging techniques of the kind we have developed during the last years. In this paper we further investigate the robustness of our CS HS techniques for the task of object classification by deep learning methods. We show preliminary results demonstrating that deep neural network classifiers perform equally well when applied on HS data captured with our compressively sensed methods, as when applied on conventionally sensed HS data.