Compressed Imaging at Long Range in SWIR

In this paper, we present a single pixel camera operating in the Short Wave InfraRed (SWIR) spectral range that reconstructs high resolution images from an ensemble of compressed measurements. The SWIR spectrum provides significant benefits in many applications due to its night vision characteristics and its ability to penetrate smoke and fog. Walsh-Hadamard matrices are used for generating pseudo-random measurements which speed up the reconstruction and enables reconstruction of high resolution images. Total variation regularization is used for finding a sparse solution in the gradient space. The edge response for the single pixel camera is analysed. A large number of outdoor scenes with varying illumination has been collected using the single pixel sensor. Visual inspection of the reconstructed SWIR images indicates that most scenes and objects can be identified after a sample ratio of 3%. The reconstruction quality improves in general as the sample ratio increases, but the quality is not improved significantly after the sample ratio has reached roughly 10%. Dynamic scenes in the form of global illumination variations can be handled by temporal local average suppression of the measurements.