Irregular Eigenimage Pyramids and Robust Appearance- Based Object Recognition (bibtex)
by Georg Langs, Horst Bischof, Walter G. Kropatsch
Abstract:
In this report we deal with the possibilities to improve the performance of robust PCA algorithms by replacing uniform subsampling of images by an irregular image pyramid. The image pyramid is built based on knowledge gained from the training set of images. It represents different regions of the image with different level of detail, depending on their importance for the reconstruction process. As this is a technical report, not only the final and optimal results will be presented, but also some of the previous steps will be explained, even if they were abandoned later. The main reason is to support future work on the topic.
Reference:
Irregular Eigenimage Pyramids and Robust Appearance- Based Object Recognition (Georg Langs, Horst Bischof, Walter G. Kropatsch), Technical report, PRIP, TU Wien, 2002.
Bibtex Entry:
@TechReport{TR067,
  author =	 "Georg Langs and Horst Bischof and Walter
                  G. Kropatsch",
  institution =	 "PRIP, TU Wien",
  number =	 "PRIP-TR-067",
  title =	 "Irregular {E}igenimage {P}yramids and {R}obust
                  {A}ppearance- {B}ased {O}bject {R}ecognition",
  year =	 "2002",
  url =		 "ftp://ftp.prip.tuwien.ac.at/pub/publications/trs/tr67.pdf",
  abstract =	 "In this report we deal with the possibilities to
                  improve the performance of robust PCA algorithms by
                  replacing uniform subsampling of images by an
                  irregular image pyramid. The image pyramid is built
                  based on knowledge gained from the training set of
                  images. It represents different regions of the image
                  with different level of detail, depending on their
                  importance for the reconstruction process. As this
                  is a technical report, not only the final and
                  optimal results will be presented, but also some of
                  the previous steps will be explained, even if they
                  were abandoned later. The main reason is to support
                  future work on the topic.",
}
Powered by bibtexbrowser