Christoph C. Borel

Recent advances in temperature-emissivity separation algorithms

This paper presents recent advances in temperature emissivity separation. 12 The ARTEMISS (Automatic Retrieval of Temperature and Emissivity using Spectral Smoothness) method was developed over the last 10 years to retrieve temperature, emissivity and associated uncertainties from data collected with hyper-spectral imaging spectrometers. Numerous improvements have been made to speed up the algorithm and create better results. In particular, new selection methods to find blackbody-like pixels led to better estimates of transmission and path radiances. New methods to select the best atmosphere, based on transmission and path radiance, narrow down the number of possible solutions. A new multi-dimensional fitting method based on the AMOEBA algorithm made it possible to include changes on the down-welling radiance for airborne sensors, e.g. from clouds and from transmission through sub-visible cirrus clouds.

Improving the detectability of small spectral targets through spatial filtering

In this paper we discuss our approach to winning entries to the RIT blind test competition. The image cube was preprocessed using a spatial filter that changed the sharpness and enhanced and isolated small point like features. This spatially sharpened cube was then processed using the ENVI hour glass algorithm and obtained high probability of detection and a small probability of false alarm for the blind test targets. In a simulation we quantified this result using metrics related to the Receiver Operator Characteristics (ROC) curve analysis. A hyper-spectral data cube was created and sub-pixel targets were inserted. We found that sharpening the hyper-spectral cube increases the number of correctly identified sub-pixel targets compared to no pre-processing. In particular the simple un-sharp masking filter generates excellent results. We propose that all sub-pixel target detection algorithms could benefit from sharpening of the spectral cube.

Simulating systematic scene-change artifacts in Fourier-transform spectroscopy

Improved understanding of midwave infrared (1-5μm) spectral emissions from detonation fireballs is needed to develop a battle space optical forensics capability. While Fourier-transform spectrometers (FTS) are an attractive tool, interferometer-based spectroscopic measurements can be corrupted when the observed scene intensity systematically varies during the measurement time. Approximating a detonation fireball as a blackbody radiator with a time-varying temperature T and modified by atmospheric attenuation τ(~ν), double-sided interferograms from an ideal FTS were calculated and converted to measured spectra Lm(~ν) to characterize the nature and magnitude of scene-change artifacts. T(x) decreased exponentially with optical path difference x, -xm ≤ x ≤ xm, at various rates relative to the Michelson mirror speed so that changing scene spectra could be simulated on 1700 ≤ ~ν ≤ 7900cm-1 at δ ~ν = 3.64cm-1 resolution (xm = 0.25cm, Hamming apodization). The real part of Lm(~ν), Re{Lm(~ν)}, is well approximated by the instantaneous spectrum at zero path difference, L(~ν,x = 0). In regions where τ(~ν) is highly structured, both the imaginary component Im{Lm(~ν)} and the differences between Re{Lm(~ν)} and L(~ν,0) exhibit spectral features, and in general |Im{Lm(~ν)}|>>|Re{Lm(~ν)}-L(~ν,0)|. In a region of highly structured absorption, 2800 ≤~ν ≤ 3500cm-1, a 600K decrease in temperature produced RMS values of 62 and 5μW/(cm2 • sr •cm-1) in Im{Lm(~ν)} and Re{Lm( ~ν)-L(~ν,0)}, respectively, compared with an RMS value of 1924μW/ (cm2 • sr • cm-1) in Re{Lm(~ν)}. A method based on theoretical expressions developed by Kick et al. is devised to interpret Lm(~ν) and provide estimates of the temporal evolution T(x) when its functional formis not known a priori.

Range-invariant anomaly detection applied to imaging Fourier transform spectrometry data

This paper describes the end-to-end processing of image Fourier transform spectrometry data taken of surrogate tank targets at Picatinny Arsenal in New Jersey with the long-wave hyper-spectral camera HyperCam from Telops. The first part of the paper discusses the processing from raw data to calibrated radiance and emissivity data. The second part discusses the application of a range-invariant anomaly detection approach to calibrated radiance, emissivity and brightness temperature data for different spatial resolutions and compares it to the Reed-Xiaoli detector.

Data processing and temperature-emissivity separation for tower-based imaging Fourier transform spectrometer data

In this article we describe the end-to-end processing of image Fourier transform spectrometry data taken at Picatinny Arsenal in New Jersey with the long-wave hyperspectral camera manufactured by the Telops company. The first part of the article discusses the processing from raw data to calibrated radiance and emissivity data. Data were taken during several months under different weather conditions every 6 min from a 213 ft-high tower of surrogate tank targets for a project sponsored by the Army Research Laboratory in Adelphi, Maryland. The second part discusses environmental effects such as diurnal and seasonal atmospheric and temperature changes and the effect of cloud cover on the data. To test the effect of environmental conditions, a range-invariant anomaly detection approach is applied to calibrate radiance, brightness temperature, and emissivity data. The data set presented in this article is due to be released publicly so that more detailed studies can be performed not only on the man-made targets but also on the changes of natural background of vegetation and gravel with time of day and seasons.

Methods to find sub-pixel targets in hyperspectral data

This paper investigates the improvement in sub-pixel target detection when image sharpening is applied to the data. A hyperspectral data cube was created using random linear mixtures of spectra and a grid of sub-pixel targets were inserted. The data cube was then convolved with a point-spread function to simulate blurring, noise was added and the output quantized. The resulting image cube is then pre-processed using various sharpening algorithms. We found that sharpening the hyperspectral cube generally increases the number of correctly identified sub-pixel targets compared to no pre-processing. In a simulation we quantified this result using a clutter matched filter ratio. We propose that all sub-pixel target detection algorithms could benefit from sharpening of the spectral cube.

Novel methods for panchromatic sharpening of multi/hyper-spectral image data

In this paper we address six problems we have encountered when sharpening multi-spectral imagery (MSI) using panchromatic (PAN) images and describe methods we have developed to solve them. We also describe a PAN-sharpening method that can be used for hyper-spectral data where the PAN-band does not cover all spectral bands. We also compare a number of currently used PAN-sharpening methods. The comparison is done by visually creating true and false color composites and by computing their radiometric fidelity with the Wang-Bovik quality index.

Algorithm for retrieving vegetative canopy and leaf parameters from multi- and hyperspectral imagery

In recent years hyper-spectral data has been used to retrieve information about vegetative canopies such as leaf area index and canopy water content. For the environmental scientist these two parameters are valuable, but there is potentially more information to be gained as high spatial resolution data becomes available. We developed an Amoeba (Nelder-Mead or Simplex) based program to invert a vegetative canopy radiosity model coupled with a leaf (PROSPECT5) reflectance model and modeled for the background reflectance (e.g. soil, water, leaf litter) to a measured reflectance spectrum. The PROSPECT5 leaf model has five parameters: leaf structure parameter Nstru, chlorophyll a+b concentration Cab, carotenoids content Car, equivalent water thickness Cw and dry matter content Cm. The canopy model has two parameters: total leaf area index (LAI) and number of layers. The background reflectance model is either a single reflectance spectrum from a spectral library() derived from a bare area pixel on an image or a linear mixture of soil spectra. We summarize the radiosity model of a layered canopy and give references to the leaf/needle models. The method is then tested on simulated and measured data. We investigate the uniqueness, limitations and accuracy of the retrieved parameters on canopy parameters (low, medium and high leaf area index) spectral resolution (32 to 211 band hyperspectral), sensor noise and initial conditions.

Against conventional wisdom: Longitudinal inference for pattern recognition in remote sensing

In response to Democratization of Imagery, a recent leading theme in the scientific community, we discuss a persistent imaging experiment dataset, which is being considered for public release in a foreseeable future, and present our observations analyzing a subset of the dataset. The experiment is a long-term collaborative effort among the Army Research Laboratory, Army Armament RDEC, and Air Force Institute of Technology that focuses on the collection and exploitation of longwave infrared (LWIR) hyperspectral and polarimetric imagery. In this paper, we emphasize the inherent challenges associated with using remotely sensed LWIR hyperspectral imagery for material recognition, and argue that the idealized data assumptions often made by the state of the art methods are too restrictive for real operational scenarios. We treat LWIR hyperspectral imagery for the first time as Longitudinal Data and aim at proposing a more realistic framework for material recognition as a function of spectral evolution over time. The defining characteristic of a longitudinal study is that objects are measured repeatedly through time and, as a result, data are dependent. This is in contrast to cross-sectional studies in which the outcomes of a specific event are observed by randomly sampling from a large population of relevant objects, where data are assumed independent. The scientific community generally assumes the problem of object recognition to be cross-sectional. We argue that, as data evolve over a full diurnal cycle, pattern recognition problems are longitudinal in nature and that by applying this knowledge it may lead to better algorithms.

Simulation of sub-pixel thermal target detection

Spectral unmixing is commonly used in the solar-reflective spectral regime to find materials down to abundances (‘pixel fill factors’ (PFFs) or ‘target fractions’) of (typically) ∼10%.12 However, little is known about whether such spectral unmixing applies in the long-wavelength-infrared (LWIR) spectral regime (designated herein as the “thermal” region). The first part of this paper discusses a numerical method to un-mix thermal pixels for target fraction, temperature and emissivity (ε). In the second part, a simulation is performed to detect particular material spectra at the sub-pixel level. The data is assumed to have been reduced to a temperature map and an emissivity data cube (a three-dimensional construct consisting of emissivity versus pixel position in the image plane corresponding to a given spectral interval; referred to herein as a ‘cube’ consisting of ε, x, y). Thus, a hyperspectral emissivity cube is generated and sub-pixel targets are added. The scene is then blurred. Various pre-processing algorithms are then evaluated with respect to the detectability of sub-pixel targets. The result of sharpening a hyperspectral cube increases the number of correctly identified sub-pixel targets compared to attempts at target detection with no pre-processing. In particular, the simple sharpened masking filter generates excellent results. We believe that many sub-pixel target detection algorithms could potentially benefit from sharpening of the spectral data cube.