Spatial-spectral feature extraction for hyperspectral image classification

Jie Liang (ANU)

COMPUTER VISION AND ROBOTICS SERIES

DATE: 2013-11-21
TIME: 15:00:00 - 16:00:00
LOCATION: NICTA - 7 London Circuit
CONTACT: JavaScript must be enabled to display this email address.

ABSTRACT:
In traditional computer area, image classification mainly relies on imageas intensity, color and texture. For remote hyperspectral images, its spectral information is the main consideration with tens or hundreds continuous light wavelength indexed bands. With hyperspectral imaging coming into computer vision area, how to take use of the extra spectral information and deal with the high dimensionality problem is still challenging. My PhD topic will focus on the spatial-spectral feature extraction on hyperspectral images. Efforts will be put at the combination of spatial features in tradition computer vision area and spectral feature in remote sensing. This new type of features will be tested in image classification. In my talk, I will give an introduction and my research plan on this topic.
BIO: