By Topic

A hierarchical boosting algorithm based on feature selection for Synthetic Aperture Radar image retrieval

Sign In

Cookies must be enabled to login.After enabling cookies , please use refresh or reload or ctrl+f5 on the browser for the login options.

Formats Non-Member Member
$33 $13
Learn how you can qualify for the best price for this item!
Become an IEEE Member or Subscribe to
IEEE Xplore for exclusive pricing!
close button

puzzle piece

IEEE membership options for an individual and IEEE Xplore subscriptions for an organization offer the most affordable access to essential journal articles, conference papers, standards, eBooks, and eLearning courses.

Learn more about:

IEEE membership

IEEE Xplore subscriptions

4 Author(s)
Mengling Liu ; Signal Processing Lab., Electronic Information School, Wuhan University, 430079, China ; Chu He ; Chao Qian ; Hong Sun

A hierarchical boosting algorithm based on feature selection is proposed for Synthetic Aperture Radar (SAR) image retrieval here. Motivated by Joint Boost and Shared feature frameworks, category combinations are adopted as the training and classification set of a hierarchical boosting-based classification frameworkpsilas middle layer. It has superiorities over the classical method which combines Boosting algorithm with many features as inputs. Meanwhile, different from the Joint Boost scheme, our method separates feature selection from training and retrieval processes. Thus more flexible feature selecting schemes can be used, e.g. nonlinear separating plane can be obtained. Some typical features such as Gabor, Edge Orientation Histogram, gray-level co-occurrence matrix, Grey Histogram and Tamura are used as the candidates of the input and statistics-based selecting method is used as the feature selection scheme. The experiments are carried on the KTH_TIPS and SAR image datasets and the results reveal our algorithmpsilas efficient performances and superiorities.

Published in:

2008 9th International Conference on Signal Processing

Date of Conference:

26-29 Oct. 2008