A technique of classifying remote sensing images based on hybrid approach

Abstract

Remote sensing image classification is interested by reseachers. Having two main approachs include the pixel based approach and the
object oriented based approach. Remote sensing images can have multichannel and high resolution. The pixel based approach usually has high accuracy but having problem with large size images as remote sensing images. While the object oriented based approach the problem but usually having accuracy lower than the pixel based approach. This paper presents a hybrid technique which combines both of the approachs to propose a more effective classifying algorithm of the remote sensing image classification. The algorithm is experimented on
data set which is remote sensing images of Hoa Binh province.