Investigation of land use classification algorithms using images fusion techniques (Case study: Beheshtabad Sub-basin)

Subject Areas : Geospatial systems development

Arash  Fazeli Farsani ¹ (MSc. Graduated of Watershed Management, Kashan University, Kashan, Iran)
Reza  Ghazavi ² (Assoc. Prof. College of Natural Resources and Earth Sciences, Kashan University, Kashan, Iran)
Mohammad Reza  Farzaneh ³ (PhD. Student of Water Resources, Tarbiat Modares University, Tehran, Iran)

Received: 2014-04-26 Accepted : 2014-11-24 Published : 2015-05-22

Keywords:

Abstract :

In recent decades, land use changes cause an important environmental, economical and social damage in the world. Land use classification images are essential tools for land management and policy decisions.  Fusion of remote sensing data could increase quality and intensify of data. The main aim of this study was investigated of land use classification algorithms (Support Vector Machine Background, Neural Net, and Maximum Likelihood) using a fusion of OLI remote sensing data using Geographic Information System (GIS). According to the results, Gram-Schmidt method had the best punctuality for taking of the firstly data and rebuilding of images. The results showed that Maximum likelihood method increased accuracy of PC_Sharpened and Gram-Schmidt algorithms. A significant accuracy in land use classification was observed via combination techniques.