Multi-class Support Vector Machine: A new approach to characterize a texture

Our former work primarily concerned the classification of satellite images after having coded their textures by a new approach of coding. These texture characteristics are extracted using cooccurences matrix because of their wealth of information from texture. The results obtained showed the interes...

Full description

Saved in:
Bibliographic Details
Main Authors: Hanifi, M., Sedes, F., Aboutajdine, D., Lasfar, A.
Format: Conference Proceeding
Language:English
Subjects:
Data mining
Image coding
Kernel
Pattern recognition
Quadratic programming
Satellites
Statistical learning
Supervised learning
Support vector machine classification
Support vector machines
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Our former work primarily concerned the classification of satellite images after having coded their textures by a new approach of coding. These texture characteristics are extracted using cooccurences matrix because of their wealth of information from texture. The results obtained showed the interest from the second coding, which will be explained later on, and which will improve the results of the first coding. At first, we present, briefly, the first coding which reduces the number of gray levels while passing from 256 levels to 9 gray levels; this phase will serve to code original textures. Then, we show how the second coding will be makes increase the levels of gray to 16, and improves quality of the image. Lastly, classification by SVM will be carried out in order to show the performance of the method used while varying the various parameters of the SVM.
ISSN:2163-5137
DOI:10.1109/ISIE.2007.4374861