Hybrid Algorithm for Classifying Soil Types Using Hyperspectral Images

Abstract

In field survey, soil type is the key indicator, but these soil classifications are depending on the personal experience. To classify types of soil physically is difficult so hyperspectral image is used to classify soil types quickly and accurately. These hyperspectral image is classified with texture features and spectral information. In this paper types of soils are classified by applying maximum likelihood and K-Nearest Neighbour algorithms to hyperspectral image. Also a Hybrid algorithm which is a combination of Maximum Likelihood and K-Nearest Neighbour was proposed to classify the soil types and applied on HSI. The outcome indicates that hybrid algorithm is the most efficient technology for classification of various types of soils.