David Bergström

Four-parameter model for polarization-resolved rough-surface BRDF

A modeling procedure is demonstrated, which allows representation of polarization-resolved BRDF data using only four parameters: the real and imaginary parts of an effective refractive index with an added parameter taking grazing incidence absorption into account and an angular-scattering parameter determined from the BRDF measurement of a chosen angle of incidence, preferably close to normal incidence. These parameters allow accurate predictions of s- and p-polarized BRDF for a painted rough surface, over three decades of variation in BRDF magnitude. To characterize any particular surface of interest, the measurements required to determine these four parameters are the directional hemispherical reflectance (DHR) for s- and p-polarized input radiation and the BRDF at a selected angle of incidence. The DHR data describes the angular and polarization dependence, as well as providing the overall normalization constraint. The resulting model conserves energy and fulfills the reciprocity criteria.

High spatial resolution hyperspectral camera based on a linear variable filter

A small, lightweight, and inexpensive hyperspectral camera based on a linear variable filter close to the focal plane array (FPA) is described. The use of a full-frame sensor allows large coverage with high spatial resolution at moderate spectral resolution. The spatial resolution has been maintained using a tilt/shift lens for chromatic focusing corrections. The trade-offs of positioning the filter relative to the FPA and varying the f-number have been studied. Calibration can correct for artifacts such as spectral filter variability. Reference spectra can be obtained using the same camera system by imaging targets over homogeneous areas. For textured surfaces, the different materials can be separated by using statistical methods. Accurate reconstruction of the sparse spectral image data is demonstrated.

Noise properties of a corner-cube Michelson interferometer LWIR hyperspectral imager

Interferometric hyperspectral imagers using infrared focal plane array (FPA) sensors have received increasing interest within the field of security and defence. Setups are commonly based upon either the Sagnac or the Michelson configuration, where the former is usually preferred due to its mechanical robustness. However, the Michelson configuration shows advantages in larger FOV due to better vignetting performance and improved signal-to-noise ratio and cost reduction due to relaxation of beamsplitter specifications. Recently, a laboratory prototype of a more robust and easy-to-align corner-cube Michelson hyperspectral imager has been demonstrated. The prototype is based upon an uncooled bolometric FPA in the LWIR (8-14 μm) spectral band and in this paper the noise properties of this hyperspectral imager are discussed.

Hyperspectral image analysis using deep learning — A review

Deep learning is a rather new approach to machine learning that has achieved remarkable results in a large number of different image processing applications. Lately, application of deep learning to detect and classify spectral and spatio-spectral signatures in hyperspectral images has emerged. The high dimensionality of hyperspectral images and the limited amount of labelled training data makes deep learning an appealing approach for analysing hyperspectral data. Auto-Encoder can be used to learn a hierarchical feature representation using solely unlabelled data, the learnt representation can be combined with a logistic regression classifier to achieve results in-line with existing state-of-the-art methods. In this paper, we compare results between a set of available publications and find that deep learning perform in line with state-of-the-art on many data sets but little evidence exists that deep learning outperform the reference methods.

Design, calibration and characterization of a low-cost spatial Fourier transform LWIR hyperspectral imaging camera with spatial and temporal scanning modes

Hyperspectral imaging in the longwave infrared region (LWIR) offers a unique day and night sensor capability currently not available in other spectral ranges. Current proliferation of the technology is however often limited by the size, weight, power and cost (SWaP-C) requirements of the available instruments. This paper presents a low-cost spatial Fourier Transform LWIR hyperspectral imaging camera, based on a corner cube Michelson interferometer and an uncooled microbolometer. In addition to conventional spatial pushbroom scanning, e.g. provided by a moving platform, the current interferometric setup can use temporal scanning of a stationary field-of-view due to a spatially offset stepper motor controlled corner cube retroreflector. The design and calibration and a characterization of the instrument are presented. Applications and future modifications are also discussed.

Optical polarization: background and camouflage

Polarimetric imaging sensors in the electro-optical region, already military and commercially available in both the visual and infrared, show enhanced capabilities for advanced target detection and recognition. The capabilities arise due to the ability to discriminate between man-made and natural background surfaces using the polarization information of light. In the development of materials for signature management in the visible and infrared wavelength regions, different criteria need to be met to fulfil the requirements for a good camouflage against modern sensors. In conventional camouflage design, the aimed design of the surface properties of an object is to spectrally match or adapt it to a background and thereby minimizing the contrast given by a specific threat sensor. Examples will be shown from measurements of some relevant materials and how they in different ways affect the polarimetric signature. Dimensioning properties relevant in an optical camouflage from a polarimetric perspective, such as degree of polarization, the viewing or incident angle, and amount of diffuse reflection, mainly in the infrared region, will be discussed.

Three-dimensional hyperspectral imaging technique

Hyperspectral remote sensing based on unmanned airborne vehicles is a field increasing in importance. The combined functionality of simultaneous hyperspectral and geometric modeling is less developed. A configuration has been developed that enables the reconstruction of the hyperspectral three-dimensional (3D) environment. The hyperspectral camera is based on a linear variable filter and a high frame rate, high resolution camera enabling point-to-point matching and 3D reconstruction. This allows the information to be combined into a single and complete 3D hyperspectral model. In this paper, we describe the camera and illustrate capabilities and difficulties through real-world experiments.

Characterization and performance of a LWIR polarimetric imager

Polarimetric information has been shown to provide means for potentially enhancing the capacity of electro-optical sensors in areas such as target detection, recognition and identification. The potential benefit must be weighed against the added complexity of the sensor and the occurrence and robustness of polarimetric signatures. While progress in the design of novel systems for snapshot polarimetry may result in compact and lightweight polarimetric sensors, the aim of this work is to report on the design, characterization and performance of a polarimetric imager, primarily designed for polarimetric signature assessment of static scenes in the long wave thermal infrared. The system utilizes the division-of-time principle and is based on an uncooled microbolometer camera and a rotating polarizing filter. Methods for radiometric and polarimetric calibrations are discussed. A significant intrinsic polarization dependency of the microbolometer camera is demonstrated and it is shown that the ability to characterize, model and compensate for various instrument effects play a crucial role for polarimetric signature assessment.

Deep subspace mapping in hyperspectral imaging

We propose a novel Deep learning approach using autoencoders to map spectral bands to a space of lower dimensionality while preserving the information that makes it possible to discriminate different materials. Deep learning is a relatively new pattern recognition approach which has given promising result in many applications. In Deep learning a hierarchical representation of increasing level of abstraction of the features is learned. Autoencoder is an important unsupervised technique frequently used in Deep learning for extracting important properties of the data. The learned latent representation is a non-linear mapping of the original data which potentially preserve the discrimination capacity.

Uncertainty analysis of sensor performance parameters in the shortwave infrared spectral range based on nightglow as the main lightsource

Images collected in the shortwave infrared (SWIR) spectral range, 1-2.5 μm, are similar to visual (VIS) images and are easier to interpret for a human operator than images collected in the thermal infrared range, >3 μm. The ability of SWIR radiation to penetrate ordinary glass also means that conventional lens materials can be used. The night vision capability of a SWIR camera is however dependent on external light sources. At moonless conditions the dominant natural light source is nightglow, but the intensity is varying, both locally and temporally. These fluctuations are added to variations in other parameters and therefore the real performance of a SWIR camera at moonless conditions can be quite different compared with the expected performance. Collected measured data from the literature on the temporal and local variations of nightglow are presented and the variations of the nightglow intensity and other measured parameters are quantified by computing standard and combined standard uncertainties. The analysis shows that the uncertainty contributions from the nightglow variations are significant. However, nightglow is also found to be a potentially adequate light source for SWIR applications.