Exploiting Information Theory for Filtering the Kadir Scale-Saliency Detector

Authors:
Pablo Suau;Francisco Escolano
Affiliations:
Departamento de Ciencia de la Computación e Inteligencia Artificial, Universidad de Alicante, Ap. de correos 99, 03080, Alicante, Spain;Departamento de Ciencia de la Computación e Inteligencia Artificial, Universidad de Alicante, Ap. de correos 99, 03080, Alicante, Spain
Venue:
IbPRIA '07 Proceedings of the 3rd Iberian conference on Pattern Recognition and Image Analysis, Part II
Year:
2007

Citing 7
Cited 0

Elements of information theory

Elements of information theory
Saliency, Scale and Image Description

International Journal of Computer Vision
Statistical Edge Detection: Learning and Evaluating Edge Cues

IEEE Transactions on Pattern Analysis and Machine Intelligence
The Distinctiveness, Detectability, and Robustness of Local Image Features

CVPR '05 Proceedings of the 2005 IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR'05) - Volume 2 - Volume 02
A Comparison of Affine Region Detectors

International Journal of Computer Vision
Kernel-based Recognition of Human Actions Using Spatiotemporal Salient Points

CVPRW '06 Proceedings of the 2006 Conference on Computer Vision and Pattern Recognition Workshop
Two Bayesian methods for junction classification

IEEE Transactions on Image Processing

Quantified Score

Hi-index	0.00

Visualization

Abstract

In this paper we propose a Bayesian filter for the Kadir Scale Saliency Detector. Such filter is addressed to deal with the main bottleneck of the Kadir detector, which is the scale space search for all pixels in the image. Given some statistical knowledge about images considered, we show that it is possible to discard some points before applying the Kadir detector by using Information Theory and Bayesian Analysis, increasing efficiency with low error. Our method is based on the intuitive idea that homogeneous (not salient) image regions at high scales probably will be also homogeneous at lower scales of scale space.