A high-resolution SWIR camera via compressed sensing

Abstract

Images from a novel shortwave infrared (SWIR, 900 nm to 1.7 µm) camera system are presented. Custom electronics and software are combined with a digital micromirror device (DMD) and a single-element sensor; the latter are commercial off-the-shelf devices, which together create a lower-cost imaging system than is otherwise available in this wavelength regime. A compressive sensing (CS) encoding schema is applied to the DMD to modulate the light that has entered the camera. This modulated light is directed to a single-element sensor and an ensemble of measurements is collected. With the data ensemble and knowledge of the CS encoding, images are computationally reconstructed. The hardware and software combination makes it possible to create images with the resolution of the DMD while employing a substantially lower-cost sensor subsystem than would otherwise be required by the use of traditional focal plane arrays (FPAs). In addition to the basic camera architecture, we also discuss a technique that uses the adaptive functionality of the DMD to search and identify regions of interest. We demonstrate adaptive CS in solar exclusion experiments where bright pixels, which would otherwise reduce dynamic range in the images, are automatically removed.

Why Sparse? Fuzzy Techniques Explain Empirical Efficiency of Sparsity-Based Data- and Image-Processing Algorithms

“…In many practical applications, it turned out to be efficient to assume that the signal or an image is sparse; see, e.g., [1], [2], [3], [4], [5], [6], [7], [8], [9], [12], [13], [14], [17], [18].…”

Characterization of a compressive imaging system using laboratory and natural light scenes

“…To characterize single photodetector CI system performance as a function of encoding basis sets, we built the test-bed system similar to those of previous works [4][5][6][7] (shown schematically in Fig. 1).…”

“…Even without a priori knowledge of the scene, perfect reconstruction can be possible with sampling below the Nyquist limit [1][2][3]. Compressive imaging (CI) systems acquire scene information through a series of measurements of the total energy transmitted through image plane spatial filters (basis functions), using the minimum number of measurements required to achieve the desired image quality or, more generally, the feature-based scene information [4][5][6][7]. CI systems are of great potential interest for specific applications where focal planes with the desired spectral or spatial resolution are unavailable, or where adaptive feature-based sensing may operate with few measurements.…”

High-resolution imaging using a translating coded aperture

“…Olivas et al 10 have evaluated the performance of a DMD-based CI system using multiple basis sets on monochromatic, color, and infrared natural light scenes, and compared the results to a camera with the same number of (conventional pixel) measurements. McMackin et al 11 have described a compressive short-wave infrared (SWIR) imager (900 nm to 1.7 μm) camera with a DMD and a single-element sensor. They showed that the hardware and software combination makes it possible to create images with the resolution of the DMD while employing a substantially lower cost sensor subsystem than would otherwise be required by the use of traditional FPAs.…”