Contextual Detection of Anomalies within Hyperspectral Images

Abstract

The majority of anomaly detectors in Hyperspectral Imaging use only the statistical aspects of the spectral readings in the image. These detectors fail to use the spatial context that is contained in the images. The use of this information can yield detectors that out perform their spatially myopic counterparts. To demonstrate this, we will use an independent component analysis based detector, autonomous global anomaly detector (AutoGAD), developed at AFIT augmented to account for the spatial context of the detected anomalies. Through the use of segmentation algorithms, the anomalies identified are formed into regions. The size and shape of these regions are then used to decide if the region is anomalous or not. A Bayesian Belief Network structure is used to update a posterior probability of the region being anomalous.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Mar 01, 2011
Accession Number
ADA540981

Entities

People

  • Adam J. Messer

Organizations

  • Air Force Institute of Technology

Tags

Communities of Interest

  • Autonomy
  • Energy and Power Technologies
  • Sensors

DTIC Thesaurus Topics

  • Air Force
  • Analysis Of Variance
  • Anomaly Detection
  • Bayesian Networks
  • Data Mining
  • Detection
  • Detectors
  • Electromagnetic Spectra
  • Hyperspectral Imagery
  • Image Processing
  • Information Science
  • Machine Learning
  • Operations Research
  • Probability
  • Remote Sensing
  • Scattering
  • Signal Processing

Fields of Study

  • Physics

Readers

  • Image Processing and Computer Vision.
  • Neural Network Machine Learning.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference