Combining Visible and Near Infrared Cues for image Categorisation
Neda Salamati, Diane Larlus, Gabriela Csurka
In this paper, we consider the scene classification problem in the context of images for which both standard visible RGB channels and near infrared information are available.
Using a popular and well performing local patch based image representation, we
propose a thorough experimental on this new category of data. We determine which
descriptors are relevant, and how to best combine them. We show that combining both
texture and color information computed on color and near infrared cues by late fusion produces results that outperform state of the art.
BMVC, Dundee, UK, August 29- September 2, 2011.
2011-025bmvc_.pdf (5.22 MB)