Content-Based Image Orientation Detection
with Support Vector Machines
Yongmei Wang
Department of Information Engineering
The Chinese University of Hong Kong
and
Hongjiang Zhang
Media Computing Group
Microsoft Research Asia
IEEE Content-Based Access of Image and Video Libraries,
pages 17-23, December 2001
Abstract
Accurate and automatic image orientation detection is of great importance in
image libraries. In this paper, we present automatic image orientation detection
algorithms by adopting both the illuminance (structural) and chrominance (color)
low-level content features. The statistical learning Support Vector Machines
(SVMs) are used in our approach as the classifiers. The different sources of
the extracted image features, as well as the binary classification nature of
SVM, require our system to be able to integrate the outputs from multiple
classifiers. Both static combiner (averaging) and trainable combiner (also
based on SVMs) are proposed and evaluated in this work. In addition, two
rejection options (regular and re-enforced ambiguity rejections) are employed
to improve orientation detection accuracy by sieving out images with low
confidence values during the classification. A number of experiments on a
database of more than 14,000 images were performed to validate our approaches.
back to Home page