An Overview of Nonlinear Mixing Models for Hyperspectral Imagery

Share

Multiple mixing models for hyperspectral imagery have been developed over the years. These mixing models allow researchers to extract sub-pixel information from hyperspectral imagery leading to novel detection and classification applications. The most common mixing model is the linear mixing model, which states that each pixel’s spectral signature is a linear combination of the unique materials (or endmembers) in the scene. For mixtures of particulates, an intimate mixture model exists that nonlinearly combines the spectra. For mixtures where multiple paths are involved, such as in tree canopies, a bilinear mixture model exists. Recently, new mixing models have emerged which combine different aspects of the aforementioned models. This talk will provide a description of these different mixing models as well as examples of each model applied to both in-lab and remotely sensed hyperspectral data



  Date and Time

  Location

  Hosts

  Registration



  • 54 Lomb Memorial Drive
  • Rochester, New York
  • United States 14623
  • Building: Center for Imaging Science
  • Room Number: 1125 Aud.
  • Click here for Map

Staticmap?size=250x200&sensor=false&zoom=14&markers=43.0858914%2c 77


  Speakers

Joshua Broadwater of Applied Physics Laboratory

Topic:

An Overview of Nonlinear Mixing Models for Hyperspectral Imagery

Multiple mixing models for hyperspectral imagery have been developed over the years. These mixing models allow researchers to extract sub-pixel information from hyperspectral imagery leading to novel detection and classification applications. The most common mixing model is the linear mixing model, which states that each pixel’s spectral signature is a linear combination of the unique materials (or endmembers) in the scene. For mixtures of particulates, an intimate mixture model exists that nonlinearly combines the spectra. For mixtures where multiple paths are involved, such as in tree canopies, a bilinear mixture model exists. Recently, new mixing models have emerged which combine different aspects of the aforementioned models. This talk will provide a description of these different mixing models as well as examples of each model applied to both in-lab and remotely sensed hyperspectral data

Biography:

Dr. Broadwater received his B.Eng. degree with double majors in Electrical Engineering and Applied Mathematics from Vanderbilt University, Nashville, TN, USA in 1994; his M.S.E.E. degree from The Georgia Institute of Technology, Atlanta, GA, USA in 1996; and his Ph.D. in Electrical Engineering from the University of Maryland, College Park, MD, USA in 2007. He is a Principal Remote Sensing Scientist and Assistant Group Supervisor for the Image Exploitation Group at The Johns Hopkins University Applied Physics Laboratory (JHU/APL) in Laurel, MD. His research centers on adaptive and semi-supervised classification methods, incorporation of physics models into pattern recognition systems, and statistical signal processing techniques for EO/IR remote sensing systems, with a focus on hyperspectral imaging

Address:11100 Johns Hopkins Rd, , Laurel, Maryland, United States, 20723





Agenda

Seminar talk in the Center for Imaging Science at RIT.