skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Cultural Artifact Detection in Long Wave Infrared Imagery.

Abstract

Detection of cultural artifacts from airborne remotely sensed data is an important task in the context of on-site inspections. Airborne artifact detection can reduce the size of the search area the ground based inspection team must visit, thereby improving the efficiency of the inspection process. This report details two algorithms for detection of cultural artifacts in aerial long wave infrared imagery. The first algorithm creates an explicit model for cultural artifacts, and finds data that fits the model. The second algorithm creates a model of the background and finds data that does not fit the model. Both algorithms are applied to orthomosaic imagery generated as part of the MSFE13 data collection campaign under the spectral technology evaluation project.

Authors:
 [1];  [1];  [1]
  1. Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)
Publication Date:
Research Org.:
Sandia National Lab. (SNL-NM), Albuquerque, NM (United States)
Sponsoring Org.:
USDOE National Nuclear Security Administration (NNSA), Office of Defense Nuclear Nonproliferation (NA-20)
OSTI Identifier:
1339493
Report Number(s):
SAND-2017-0231
650289
DOE Contract Number:
AC04-94AL85000
Resource Type:
Technical Report
Country of Publication:
United States
Language:
English
Subject:
47 OTHER INSTRUMENTATION

Citation Formats

Anderson, Dylan Zachary, Craven, Julia M., and Ramon, Eric. Cultural Artifact Detection in Long Wave Infrared Imagery.. United States: N. p., 2017. Web. doi:10.2172/1339493.
Anderson, Dylan Zachary, Craven, Julia M., & Ramon, Eric. Cultural Artifact Detection in Long Wave Infrared Imagery.. United States. doi:10.2172/1339493.
Anderson, Dylan Zachary, Craven, Julia M., and Ramon, Eric. Sun . "Cultural Artifact Detection in Long Wave Infrared Imagery.". United States. doi:10.2172/1339493. https://www.osti.gov/servlets/purl/1339493.
@article{osti_1339493,
title = {Cultural Artifact Detection in Long Wave Infrared Imagery.},
author = {Anderson, Dylan Zachary and Craven, Julia M. and Ramon, Eric},
abstractNote = {Detection of cultural artifacts from airborne remotely sensed data is an important task in the context of on-site inspections. Airborne artifact detection can reduce the size of the search area the ground based inspection team must visit, thereby improving the efficiency of the inspection process. This report details two algorithms for detection of cultural artifacts in aerial long wave infrared imagery. The first algorithm creates an explicit model for cultural artifacts, and finds data that fits the model. The second algorithm creates a model of the background and finds data that does not fit the model. Both algorithms are applied to orthomosaic imagery generated as part of the MSFE13 data collection campaign under the spectral technology evaluation project.},
doi = {10.2172/1339493},
journal = {},
number = ,
volume = ,
place = {United States},
year = {Sun Jan 01 00:00:00 EST 2017},
month = {Sun Jan 01 00:00:00 EST 2017}
}

Technical Report:

Save / Share:
  • The purpose of Task 3.b under PL02-OP211I-PD07 (CBW simulant detection) was to demonstrate the applicability of the sensor work developed under this project for chemical and biological weapons detection. To this end, the specific goal was to demonstrate the feasibility of detection of chemical agents via that of simulants (Freons) with similar spectroscopic features. This has been achieved using Freon-125 as a simulant, a tunable external cavity quantum cascade laser (ECQCL), and a Herriott cell-based sensor developed at Pacific Northwest National Laboratory (PNNL) specifically for this task. The experimentally obtained spectrum of this simulant matches that found in the Northwestmore » Infrared (NWIR) spectral library extremely well, demonstrating the ability of this technique to detect the exact shape of this feature, which in turn indicates the ability to recognize the simulant even in the presence of significant interference. It has also been demonstrated that the detected features of a typical interferent, namely water, are so different in shape and width to the simulant, that they are easily recognized and separated from such a measurement. Judging from the signal-to-noise ratio (SNR) of the experimental data obtained, the noise equivalent absorption sensitivity is estimated to be 0.5 x 10-7 to 1 x 10-6 cm-1. For the particular feature of the simulant examined in this work, this corresponds to a relative concentration of 50 to 25 parts-per-billion by volume (ppbv). The corresponding relative concentrations of other chemical targets would differ depending on the particular transition strengths, and would thus have to be scaled accordingly.« less
  • A natural or anthropogenic process often generates a signature gas plume whose chemical constituents may be identified using hyperspectral imagery. A hyperspectral image is a pixel-indexed set of spectra where each spectrum reflects the chemical constituents of the plume, the atmosphere, the bounding background surface, and instrument noise. This study explored the relationship between gas absorbance and background emissivity across the long-wave infrared (LWIR) spectrum and how they affect relative gas detection sensitivity. The physics-based model for the observed radiance shows that high gas absorbance coupled with low background emissivity at a single wavenumber results in a stronger recorded radiance.more » Two sensitivity measures were developed to predict relative probability of detection using chemical absorbance and background emissivity: one focused on a single wavenumber while another accounted for the entire spectrum. The predictive abilities of these measures were compared to synthetic image analysis. This study simulated images with 499 distinct gases at each of 6 concentrations over 6 different background surfaces with the atmosphere and level of instrument noise held constant. The Whitened Matched Filter was used to define gas detection from an image spectrum. The estimate of a chemical’s probability of detection at a given concentration over a specific background was the proportion of detections in 500 trials. Of the 499 chemicals used in the images, 276 had estimated probabilities of detection below 0.2 across all backgrounds and concentrations; these chemicals were removed from the study. For 92.8 percent of the remaining chemicals, the single channel measure correctly predicted the background over which the chemical had the largest relative probability of detection. Further, the measure which accounted for information across all wavenumbers predicted the background over which the chemical had the largest relative probability of detection for 93.3 percent of the chemicals. These results suggest that the wavenumber with largest gas absorbance has the most influence over gas detection for this data. By furthering the in-silico experimentation with higher concentrations of gases not detectable in this experiment or by standardizing the gas absorbance spectra to unit vectors, these conclusions may be confirmed and generalized to more gases. This will help simplify image acquisition planning and the identification of unknowns in field collected images.« less
  • Natural and man-made chemical processes generate gaseous plumes that may be detected by hyperspectral imaging, which produces a matrix of spectra affected by the chemical constituents of the plume, the atmosphere, the bounding background surface and instrument noise. A physics-based model of observed radiance shows that high chemical absorbance and low background emissivity result in a larger chemical signature. Using simulated hyperspectral imagery, this study investigated two metrics which exploited this relationship. The objective was to explore how well the chosen metrics predicted when a chemical would be more easily detected when comparing one background type to another. The twomore » predictor metrics correctly rank ordered the backgrounds for about 94% of the chemicals tested as compared to the background rank orders from Whitened Matched Filtering (a detection algorithm) of the simulated spectra. These results suggest that the metrics provide a reasonable summary of how the background emissivity and chemical absorbance interact to produce the at-sensor chemical signal. This study suggests that similarly effective predictors that account for more general physical conditions may be derived.« less