Paper Details: Downloads: 401
Serial Number: P1150834318
Title: Support Vector Training of HMT Generative Models for Land Cover Image Classification
Authors: Reda A. El-Khoribi
Abstract: This paper introduces a novel approach to supervised classification of multispectral images. The approach uses a new discriminative training algorithm for discrete hidden Markov tree (HMT) generative models applied to the multi-resolution ranklet transforms. System is implemented and tested on a set of Landsat 7-band images containing eight different land cover classes. Experimental results of the system show significant improvement over the baseline HMT system and give a superior performance in land cover classification.
Keywords: HMT, SVM, land cover classification, discriminative training.
Journal/Conference: International Journal of Graphics, Vision and Image Processing
Volume:
Issue:
Submission Date: 8/21/2008 12:00:00 AM
Review Date: 9/14/2008 12:00:00 AM
Publishing Date: 9/19/2008 12:00:00 AM
Article Downloads: 401
Download:

Facebook