F Sarmadi; H Ebadi; A Mohammadzadeh
Volume 7, Issue 2 , November 2015, , Pages 55-68
Abstract
Determination of land use and land cover changes is necessary for monitoring of urban growth and responsible urban planning. Remote sensing can be used as a powerful technology in land use and land cover change detection. One of the challenges in this area is to developing efficient methods for accurate ...
Read More
Determination of land use and land cover changes is necessary for monitoring of urban growth and responsible urban planning. Remote sensing can be used as a powerful technology in land use and land cover change detection. One of the challenges in this area is to developing efficient methods for accurate and highly automated change detection which can produce accurate and precise information about position and content of the changes. In this study two GeoEye images from Tehran 17th region related to 2004 and 2010 years were used. This study Proposed a method based on image context spatial features, neural networks and genetic algorithm. Six cases with direct multi-date classification approach and post classification approach were implemented and compared in the viewpoints of accuracy and runtime. Direct multi-date classification was superior in all six cases. Between six implemented cases, sixth case (proposed method of this research) was superior in the classification accuracy point of view. In this case after selecting optimized features, ANN classification was executed based on determining architecture and several times execution. Though runtime of sixth case was the highest, if accuracy is prior, it’s highly recommended.
