Skip to main content
U.S. Department of Energy
Office of Scientific and Technical Information

Discriminating Tailing Piles using VNIR Imagery and SVMs

Technical Report ·
DOI:https://doi.org/10.2172/1573148· OSTI ID:1573148
 [1];  [1];  [1]
  1. Lawrence Livermore National Laboratory (LLNL), Livermore, CA (United States)

This report describes a brief investigation into the use of multispectral satellite imagery and a machine learning technique to discriminate between the soils in two tailing piles in the U1a Complex at the Nevada National Security Site. The south pile dates from the late 1960s and the north pile dates from the early 2000s and is where recent tailings have been dumped. Our approach is to use commercially available satellite imagery of the U1a tailing piles and a Support Vector Machine (SVM) to discriminate between the soils in these piles. Experiments were performed using multispectral pixels selected from around the U1a Complex to train the support vector machine. Pixels in the scene were then classified by the SVM. Classification performance was assessed qualitatively through study of the classification imagery output by the SVM. These qualitative assessments provide the conclusions drawn by the investigation: 1. A SVM can be used to discriminate between different types of materials in satellite imagery. 2. The VNIR imagery used in this investigation does not provide the features needed by the SVM to discriminate between the two tailing piles at the U1a Complex-the soils look too similar in these spectral bands to discriminate. The remainder of the report details the experiments performed to reach these conclusions and the limitations and caveats associated with the conclusions.

Research Organization:
Lawrence Livermore National Laboratory (LLNL), Livermore, CA (United States)
Sponsoring Organization:
USDOE National Nuclear Security Administration (NNSA)
DOE Contract Number:
AC52-07NA27344
OSTI ID:
1573148
Report Number(s):
LLNL--TR-791925; 991396
Country of Publication:
United States
Language:
English

Similar Records

Utility of multispectral imaging for nuclear classification of routine clinical histopathology imagery
Journal Article · Tue Jul 10 00:00:00 EDT 2007 · BMC Cell Biology · OSTI ID:1626368

AFREET: HUMAN-INSPIRED SPATIO-SPECTRAL FEATURE CONSTRUCTION FOR IMAGE CLASSIFICATION WITH SUPPORT VECTOR MACHINES
Conference · Wed Jan 31 23:00:00 EST 2001 · OSTI ID:774619

SUPPORT VECTOR MACHINES FOR BROAD AREA FEATURE CLASSIFICATION IN REMOTELY SENSED IMAGES
Conference · Wed Feb 28 23:00:00 EST 2001 · OSTI ID:776688

Related Subjects