Volume : IV, Issue : XII, December - 2015

Preliminary Study of Incorporating Kernels with Fuzzy Classifiers

Ishuita Sengupta

Abstract :

In Remote Sensing, Land Use/Land Cover is prepared for different project planning purpose using different classification techniques. If there are mixed pixels, classification techniques should be based upon soft approach i.e. soft classification. This paper emphasizes over a preliminary approach of implementing fuzzy classification using the incorporation of kernels based methods. The kernels are used to handle non–linearity. The concept of kernels provides the basis for the development of more robust approaches to the remote sensing classification problem; it implicitly represents the mapping of the input space to the feature space. The major objective of the paper is to demonstrate the role and effect of using kernels with fuzzy classifiers. An evaluation of kernel based fuzzy algorithms has been formulated for extracting material of interest at sub–pixel level. Output images procured after applying the classification techniques using 8 Kernels with the FCM as well as using standard FCM, every output respective to the feature class has been highlighted.

Keywords :

Article: Download PDF   DOI : 10.36106/ijsr  

Cite This Article:

Ishuita Sengupta Preliminary Study of Incorporating Kernels with Fuzzy Classifiers International Journal of Scientific Research, Vol : 4, Issue : 12 December 2015

Number of Downloads : 750

References :