| | |
Summary: ALBERT-LUDWIGS-UNIVERSIT ¨AT FREIBURG
INSTITUT F ¨UR INFORMATIK
Lehrstuhl f¨ur Mustererkennung und Bildverarbeitung
Patch Based Approaches for the Recognition of Visual Object
Classes - A Survey
Internal Report 2/06
Alexandra Teynor
November, 2006
Patch Based Approaches for Visual Object
Class Recognition - a Survey
Alexandra Teynor
November 29, 2006
1 Introduction
Today, a huge amount of digital images exists on private computers, the web, in digital galleries or professional
media archives, and the mass will continue to grow. Current image databases of publishing companies already
reach many thousands of terrabytes.
Retrieving images again from these databases is difficult. Think about your own holiday picture collection:
manual search through archives is hardly feasible on home desktop computers (if at all), not to speak of
professional archives with terrabytes of image data. Manual annotation of images and then using text retrieval
techniques on the key words is no general solution either. First of all, it is very labor and with that cost
|
